The Uninspired Status Quo

Most RWA and tokenisation projects today are focused on a painfully pedestrian goal: taking existing assets and wrapping them in a blockchain wrapper for easier consumption. Yes, this is a necessary first step, and yes, someone has to do it. But the sheer lack of imagination is staggering.

Consider the audacity of companies declaring they’re making US Treasury bonds “more liquid” by bringing them on-chain. Treasury bonds. The same instruments that Circle, a company whose entire business model consists of holding dollars for USDC, keeps the majority of its reserves in precisely because they can be liquidated instantly. The same treasuries that are literally considered the benchmark for liquidity in traditional finance. If this is innovation, then we’ve set the bar embarrassingly low.

The slightly more adventurous projects are attempting to democratise access to traditionally exclusive financial instruments – venture capital, private equity, certain types of commercial real estate that retail investors could never access unless they took the plunge themselves. These efforts, while commendable steps in the right direction, still represent incomplete and uninspired thinking about the true power of asset tokenisation.

The Paperwork Behemoths

For all its flaws and obvious inefficiencies compared to programmatic tokenisation, Wall Street has proven remarkably adept at one thing: where there’s a market to be created, the traditional financial powers will find it, create it, and pump irresponsible amounts of leverage through it if given half a chance.

But these lumbering paperwork beasts recognise themselves for what they are. Every gravy train requires its corresponding paper train, a bureaucratic apparatus that has become its own economy. When operating at scale, you inevitably lose the nuance found in smaller, specialised niches, but the economies of operational efficiency make that trade worthwhile.

This paper trail has spawned its own ecosystem of beneficiaries. Few outside the industry truly understand that mortgages are a manufactured product specifically designed to service this paper train. Let that sink in for a moment: mortgages are a manufactured product, created not out of fundamental necessity but to feed a bureaucratic machine.

The Real Revolution

This mortgage example is but one of many, yet it represents the anchor point of inspiration for where tokenisation should be heading. I am far more interested in discovering and creating RWAs that exist exclusively because of the unique powers that being on-chain brings – assets that simply could not exist under our traditional financial model.

We began orbiting this thesis a year ago when discussing the benefits of tokenisation for tackling commercial contract and investment contract law. We still stand by that thesis, and indeed Sibylline is proud of the work it has supported in delivering against it. But that thesis always felt incomplete, lacking a final evolutionary step.

De-risking major capital investment is powerful, certainly, but it’s still the same working capital that was always under discussion and, in some manner, allocated in this fashion – particularly when the underlying asset is a commodity routinely extracted at scale. What truly excites me is developing and exploring asset classes that literally would not exist without the developments being made in tokenisation.

Let me be crystal clear here for the on-chain native readers: I am not talking about re-staking or other “liquid staking tokens” that exist purely because of the evolving on-chain economy. I am specifically discussing what we would traditionally consider financial instruments that, due to a myriad of reasons, simply could never exist under our traditional financial model. The evolving landscape of liquid staking and re-staking deserves its own thesis, which is forthcoming, so we shall table that discussion for now.

The Unserved Markets

The question that keeps me up at night is this: what markets are large enough to warrant servicing but remain unserved by the paperwork behemoths that are traditional institutions? These gaps exist because markets are too fragmented, because they can’t be properly modelled, or because of some third, more esoteric reason that makes them invisible or unpalatable to traditional finance.

These pockets are either entirely unserved or, if they are served, dominated by niche players that hold complete monopolies and market power. It is RWAs that attack these spaces that I find genuinely compelling.

Beyond Digitisation: Creating New Possibilities

To understand what genuine innovation in the RWA space looks like, consider how tokenisation can create entirely new financial instruments rather than merely digitising existing ones. The Bizzed Protocol exemplifies this approach by addressing a fundamental market failure: the disconnect between local market knowledge and institutional capital deployment.

Traditional finance has systematically underserved certain markets not because they lack value, but because the paperwork machinery cannot efficiently process them. When community stakeholders possess intimate knowledge of a business but lack capital, while institutional investors have capital but lack ground-truth validation, a gap emerges that traditional structures cannot bridge. This is where tokenisation creates something genuinely new – not by wrapping existing assets, but by enabling capital structures that leverage blockchain’s unique properties to solve previously intractable problems.

The innovation lies not in putting something on-chain, but in using blockchain’s capabilities to create governance structures through technology rather than legal mechanisms, to enable transparency without bureaucracy, and to align incentives in ways that paper-based systems simply cannot achieve. This is tokenisation as financial innovation, not financial digitisation.

The Technology as Enabler, Not Focus

What separates genuine innovation from tokenisation theatre is understanding that blockchain is the enabler, not the story. Most RWA projects fail because they start with the technology – “let’s put X on the blockchain” – rather than identifying genuine market failures that blockchain’s unique properties can address.

True innovation emerges when we identify markets that traditional finance cannot serve efficiently and use tokenisation to create new instruments that address these gaps. The technology should fade into the background, serving the financial innovation rather than becoming the focal point. When projects lead with “blockchain-enabled” rather than the actual problem they solve, they’ve already lost the plot.

The Inevitable and the Revolutionary

Tokenisation of almost every traditional asset will happen eventually – the market is arriving whether we’re ready or not. But without a seriously powerful moat, the large institutions will be the ones to capture this particular gravy train as well. The crypto industry’s failure to self-regulate means it must now play in the same league as traditional finance, and TradFi has significantly more experience at this game.

This isn’t to say the space won’t accomplish incredible things, but it needs to be pragmatic about the landscape it operates in. The massive M&A activity we’re witnessing – mature crypto institutions being acquired by traditional finance players – is evidence of this change. I’ve come to view this as simply the natural technological evolution of major capital markets, analogous to the transition from pink sheets to Bloomberg terminals.

But this is precisely why we must look beyond simply tokenising existing assets. I want to explore and discover what only works because of tokenisation’s unique abilities to create, manage, and sustain assets. What could never be done before? What new forms of value can we create that were previously impossible?

The Path Forward

The future of RWA tokenisation lies not in making treasury bonds marginally more accessible or creating yet another wrapper for existing assets. It lies in identifying markets that traditional finance cannot or will not serve, and using tokenisation’s unique properties to create entirely new financial instruments that address these gaps.

On this thesis, we have identified a series of portfolio companies that truly understand this distinction. These are the pioneers who recognise that the real revolution lies in creating entirely new asset classes that leverage the unique properties of blockchain technology. They understand that tokenisation enables fundamentally new forms of value creation, not just more efficient ways of doing what we’ve always done.

The sector stands at a crossroads. It can continue down the path of uninspired digitisation, competing with traditional finance on their terms and inevitably losing. Or it can embrace genuine innovation, creating new markets and new instruments that could never exist without blockchain technology. The choice should be obvious, yet somehow, most of the industry continues to choose the former.

It’s time for this sector to move beyond its current uninspired iterations and embrace the truly revolutionary potential at its fingertips. Anything less is a failure of imagination and a betrayal of the technology’s promise. The revolution isn’t in tokenising what exists – it’s in creating what couldn’t exist before.

The Changing Guard

For decades, enterprise software was a game of scale. Oracle, SAP, Salesforce – these behemoths built their empires on the fundamental premise that complex business problems required complex solutions, and complex solutions required armies of engineers. The barrier to entry wasn’t just capital; it was the sheer technical sophistication required to build anything that could compete.

That calculus has fundamentally changed.

Today, a marketing director with a laptop and a Claude subscription can prototype a custom CRM in an afternoon. A frustrated operations manager can build a workflow automation tool over a weekend. Products and solutions that previously were dominated by a mere handful of major players with defensible economies of scale are going to be atrophied by an unchecked proliferation of differentiated players that can now compete.

The democratization of software development through AI isn’t just lowering barriers – it’s obliterating them.

The Trust Gap

Here is where the narrative gets interesting. Fortune 500 companies aren’t running their mission-critical operations on software built by someone who learned to code last Tuesday. Mature, regulated, or PE target companies are not going to take on the risk of moving workloads to vibe coded software at the top of an internet search.

The proliferation of tools has created a paradox: while it’s never been easier to build software, it’s never been harder to build trust.

Consider the procurement process at any major corporation. Security audits. Compliance certifications. Reference checks. Integration requirements. SLAs. The checklist is exhaustive, and for good reason – when your software powers a billion-dollar operation, “move fast and break things” isn’t a philosophy; it’s a lawsuit.

This is where the real opportunity emerges, hidden in plain sight.

The New Power Brokers

The winners in this new landscape won’t be the builders with the best products. They’ll be the ones with the best pedigrees.

Entrepreneurial MDs of recognised companies and respected industry leaders that instead go on that journey force multiplied by a team they can trust with technical credentials that stand up to scrutiny will emerge as the winners. These aren’t your typical startup founders. They’re the former heads of innovation at Deloitte, the ex-CTOs of mid-market success stories, the industry veterans who’ve spent decades building the very relationships that matter.

They understand something fundamental: in enterprise software, you’re not selling features. You’re selling faith.

When the former VP of Digital Transformation at a Fortune 500 launches a startup, they’re not starting from zero. They’re starting with twenty years of accumulated trust, a Rolodex of decision-makers who return their calls, and most importantly, the credibility to say “I’ve been where you are, and I built this to solve our problem.”

Social Capital as Moat

High trust relationships between founders and these firms will carry an unaffordable premium. Social capital is the new moat.

Think about what this means in practice. Two companies launch competing workflow automation platforms. One is built by a brilliant engineer with superior technology. The other is built by the former Chief Digital Officer of a major consulting firm, with marginally acceptable technology but a LinkedIn full of enterprise connections.

The market will choose the latter every time.

This isn’t a bug in the system – it’s the system working as designed. Enterprise buyers aren’t optimizing for the best technology; they’re optimizing for the lowest career risk. Nobody gets fired for buying from someone they trust.

The Coming Consolidation

We’re about to witness a fascinating market dynamic. The explosion of AI-enabled builders will create thousands of point solutions, each targeting specific niches and workflows. But the distribution will concentrate around a handful of players who can navigate the enterprise maze.

These new kingmakers will aggregate solutions, not through traditional M&A, but through trust networks. They’ll become the curators of “safe” innovation, the bridges between corporate needs and startup capabilities.

Watch for the emergence of new advisory firms, incubators led by industry veterans, and rollup strategies focused not on technology synergies but on relationship capital. The playbook is already being written by those who see the opportunity.

Implications

For non-technical founders riding the AI wave, the message is clear: your product is table stakes. Your pedigree is your product.

For enterprises, this presents both opportunity and challenge. The explosion of solutions means unprecedented choice and customization possibilities. But navigating this landscape requires new frameworks for vendor evaluation that prioritize founder credibility as much as product capability.

For investors, the calculation shifts. Due diligence on founding teams takes on new dimensions. The question isn’t just “can they build?” but “can they sell to the C-suite?”

Closing Thoughts

The democratization of software development through AI is real, transformative, and irreversible. But markets, particularly enterprise markets, don’t transform overnight. They evolve through trust networks, relationship capital, and the careful navigation of risk.

The winners in this new world won’t be those who can build the fastest or even the best. They’ll be those who understand that in enterprise software, reputation is the ultimate currency, and social capital compounds faster than any technology advantage.

The revolution is here. But it will be led by insiders, not insurgents.

The Backdrop

For those who’ve had their heads buried in the sand for the past few years, the Epic versus Apple saga has been nothing short of technological theatre worthy of Greek tragedy status. Epic Games, the creator of Fortnite and the wildly successful Unreal Engine, decided to wage war against Apple’s infamous 30% “App Store tax” and their restrictive policies that have held developers in a stranglehold since the App Store’s inception.

At its core, this was a battle about whether a single player should be able to control not just access to a platform, but also dictate the terms of commerce that happen within that ecosystem. Apple’s walled garden, beautiful as it may be to some, was deemed to be not just a garden, but a prison for developers and consumers alike.

First Order Chaos

The immediate fallout of Epic’s victory is sending shockwaves through both Wall Street and Silicon Valley. Apple’s stock is likely to take a significant hit — potentially shaving billions off its market cap — as investors recalibrate expectations for one of its most profitable revenue streams. The App Store has been a golden goose for Apple, estimated to generate upwards of $20 billion annually with profit margins that would make most businesses weep with envy.

For Epic, this is vindication of their strategy, but perhaps more importantly, it’s a catalyst for their broader ambitions. Epic’s own digital storefront could now theoretically operate on iOS devices, a previously unimaginable scenario. This alone stands to shift considerable value away from Apple and towards Epic and similar competitors.

The Second Wave: App Store Economics Decimated

The second-order effects are where things get really interesting. The App Store as we know it is, frankly, dead. The 30% commission structure that has been the backbone of Apple’s services revenue is now on life support. They’ll likely pivot to a lower fee structure — perhaps 15% across the board or even lower — to stave off mass exodus of developers to alternative payment solutions.

But it goes much deeper than just fee structures. The very concept of app review and the bottleneck it creates is now in jeopardy. If alternative app distribution channels become viable on iOS, the entire premise of Apple’s curated experience comes into question. Security, privacy, and quality control — all hallmarks of the Apple ecosystem — could be compromised in this brave new world.

The ripple effects will touch every part of the mobile ecosystem. Google, watching from the sidelines with bated breath, is next on the chopping block. Their Play Store practices, while marginally less restrictive than Apple’s, are fundamentally cut from the same cloth.

DTC Renaissance: The Third Wave Cometh

Now we get to what I believe is the most fascinating consequence: the potential explosion of direct-to-consumer (DTC) apps. For years, businesses have been forced to factor in the “Apple tax” when considering their mobile strategy. Many simply couldn’t make the economics work, particularly those with low margins or subscription-based models.

Imagine a world where Spotify doesn’t have to inflate their iOS prices or push users to sign up on their website. Imagine Netflix being able to offer in-app purchases without surrendering nearly a third of their revenue. The possibilities are staggering.

This isn’t just about established players saving money. It’s about entire business models that were previously non-viable suddenly becoming possible. Creators, small businesses, and niche services that couldn’t afford the 30% cut can now potentially thrive in a direct relationship with their customers.

The DTC revolution that’s been brewing in other sectors will finally hit mobile full force. Brands that have built direct relationships with customers through web and email can now extend that relationship to the most personal device most people own — their phone — without an intermediary taking a massive cut.

Development Paradigm Shifts

The technical ramifications of this ruling cannot be overstated. React Native, Flutter, and other cross-platform development frameworks stand to see explosive growth. With the economic incentives for platform-specific development diminishing (no need to optimize for Apple’s review process or payment systems), many businesses will pivot to solutions that allow them to build once and deploy everywhere.

The irony here is palpable. After years of Apple pushing developers toward platform-specific code with Swift and SwiftUI, they may have inadvertently accelerated the move toward platform-agnostic development. Why invest in platform-specific expertise when the economic benefits of that investment are evaporating?

The demand for React Native developers specifically will likely skyrocket. Companies looking to rapidly adjust to this new paradigm will need developers who can help them pivot quickly, and React Native’s combination of web-like development patterns and native performance makes it particularly well-suited for this transition period.

Market Concentration: The Counterintuitive Fourth Wave

Here’s where things get truly fascinating. One might assume that breaking Apple’s monopolistic grip would lead to a more diverse, competitive ecosystem. I believe the reality could be quite the opposite.

The removal of App Store restrictions could actually accelerate concentration in many app categories. Consider this: without the 30% tax, established players with strong brands and customer relationships have even more leverage. Spotify, Netflix, Amazon, and others now have both their existing scale advantages AND freedom from platform fees.

For smaller developers, the App Store at least provided a somewhat level playing field for discovery. In a world of direct distribution, those with the largest marketing budgets and existing customer bases have a tremendous advantage. The “meritocracy” of the App Store rankings, flawed as it may be, at least gave newcomers a shot.

We might see the emergence of new app store aggregators — perhaps from Epic itself, or from unexpected players like Facebook or Amazon — that become the new gatekeepers. Meet the new boss, same as the old boss.

The Long Tail: Regulatory Ripple Effects

The most enduring impact of this ruling may be on the regulatory landscape. This verdict provides a blueprint for antitrust action against other tech giants. Amazon’s marketplace, Google’s search dominance, and Facebook’s social media empire are all built on similar foundations of platform control.

Regulators worldwide, who have been struggling to find effective frameworks for reining in Big Tech, now have a precedent to build upon. The EU, already more aggressive than the US in this arena, will likely accelerate their efforts.

The irony is that this ruling, allegedly in the name of competition, may actually push the tech industry toward a more European model of regulation — with explicit rules about platform ownership and operation rather than the historically laissez-faire American approach.

Closing Remarks

The ramifications of Epic’s victory will echo far beyond the companies directly involved. We are witnessing nothing less than the restructuring of the digital economy. The market has clearly signalled that it no longer accepts the platform monopolies established in the early days of the smartphone revolution.

For developers, this is simultaneously the most exciting and terrifying shift in mobile since the introduction of the App Store itself. The opportunities are vast, but so too are the uncertainties. Business models that have been built entirely around platform economics must now be reimagined.

For consumers, this could herald an era of innovation and direct relationships with the brands and services they love. Or it could create a fragmented, confusing landscape where security and consistency are sacrificed on the altar of competition.

What’s clear is that the digital frontier is once again wide open. The rules that have governed the mobile ecosystem for over a decade have been shattered, and we are now in uncharted territory. Those who can navigate this new landscape — understanding both the technical and business implications — stand to reap extraordinary rewards.

The only certainty is change, and that change is coming faster than anyone in Cupertino would like to admit.

A guest essay by Sibylline Product Lead, Cosmo Lindsay

My first introduction to the art world came through a regional auction house in the English countryside. As a child, I was fascinated by the ritual of it all—how centuries-old paintings changed hands through subtle gestures, how tradition seemed embedded in every aspect of the process. This early curiosity eventually led me to study history of art, a path that soon intertwined with work in the art world, an industry governed by established protocols and hierarchies: meticulous provenance research, careful handling of artefacts, and reverence for institutional knowledge.

Today, I lead AI product development at Sibylline Labs in a field characterised by rapid iteration, constant innovation, and willingness to disrupt established systems. The contrast couldn’t be more striking. This transition reflects a broader phenomenon happening across industries, where people from diverse educational backgrounds—arts, humanities, social sciences—are finding their perspectives uniquely valuable in technology fields once considered inaccessible without specific technical training. The journey between these seemingly disparate domains reveals exciting insights about how we create, value, and exchange both information and creative expression in our rapidly evolving digital landscape.

The Gilded Cage

The traditional art world operates as a magnificent anachronism. Behind the glamourous private views and gleaming gallery walls lies an industry stubbornly resistant to innovation. For many starting out their early careers in the art world, they navigate a system where relationships are currency and information asymmetry is the business model. While studying art and its many movements across history, those working in the field can often find themselves embedded in perhaps the least revolutionary market structure imaginable. It’s a conversation I’ve heard time and time again.

The irony is striking: analysing Renaissance patrons’ relationships with emerging technologies while entering data into systems that predate Windows 98. This growing discord becomes increasingly difficult to reconcile; how could the keepers of creative history be so reluctant to embrace the future?

What most outside observers fail to grasp is that the art world’s resistance to technological advancement isn’t accidental, it’s structural. The white-glove service, the perfection-obsessed façade, and the deliberate opacity of pricing all serve to maintain exclusivity. In the art world, mistakes, particularly public-facing ones, aren’t merely inconveniences but existential threats to carefully cultivated images of infallibility.

My Block Zero

My first encounter with blockchain wasn’t particularly noble. In 2016, like many in my generation, I discovered Bitcoin through friends who found it a remarkably efficient tool for procuring fake IDs from the darker corners of the internet. At roughly $500 per BTC at the time, those digital transactions have since theoretically appreciated into absurdly valuable identity documents. It would be revisionist history to claim that I immediately recognised blockchain as the future of global exchange; my introduction to it came not through any great philosophical awakening but rather through witnessing its most practical, if questionable, applications.

Yet as crypto matured—weathering waves of scepticism, market volatility, and regulatory uncertainty—I found myself increasingly drawn to its architectures and use cases. It was a parallel realm operating on completely opposite principles to the rich traditions of the art market. Here was a system predicated on transparency rather than opacity, on immutable verification rather than malleable provenance, on democratised access rather than calculated gatekeeping. The conceptual juxtaposition to my own line of work was both jarring and exhilarating.

My friends from art history and creative backgrounds who ventured into crypto or NFTs experienced this same whiplash between these two seemingly antithetical worlds. The art world moved with the deliberate patience of geological time, each decision weighted by centuries of precedent and institutional memory. Meanwhile, blockchain culture hurtled forward with dizzying velocity, where a week-old protocol was considered antiquated and three-month projections qualified as “long-term planning.” Each week brought about a new hype-cycle, a new meme, a new direction.

When NFTs began to emerge on the global scene around 2020, the collision of these worlds was—to me at least—a lifeline. The technology promised to solve the very inefficiencies witnessed daily in traditional art markets: the obscure pricing mechanisms, the byzantine transaction processes, the artificial information scarcity. The promise was revolutionary, from democratised art ownership with transparent provenance tracking to the heralded idea of direct artist-to-collector relationships. The reality was messier—market manipulation, critical misunderstanding, and institutional resistance. Standing at this intersection at times felt like witnessing the early days of photography, a technology dismissed as a fad by traditional artists whilst it silently revolutionised visual culture. As Walter Benjamin noted in his essay on mechanical reproduction, new technologies don’t merely change how art is distributed, they transform how society perceives and values creative work itself, often triggering resistance from those most invested in existing systems.

The AI Inflection

My exploration of NFTs led to an unexpected detour that has coincidentally come full circle in my latest career trajectory. While working with leading artist and curator Robert Alice, I gained access to the technology behind his iNFT AI-powered avatar, an artwork that utilised GPT-3 some eighteen months before ChatGPT would capture global attention. The sense of wonder of experiencing this ‘early’ AI system generating Shakespearean sonnets on command, a technological parlour trick that hinted at something profoundly transformative, is a feeling now shared by hundreds of millions worldwide.

Managing On NFTs with TASCHEN—the world’s largest publication on the subject—provided a front-row seat to the creative side of blockchain and many of the incredible artists operating in that domain. For two and a half years, living at the bleeding edge revealed something crucial: the traditional art world and emerging technologies weren’t merely different industries but different epistemologies—distinct ways of understanding how culture evolves, how value is created, and how human creativity manifests.

This collision triggered unprecedented acceleration from typically archaic institutions. Auction houses rapidly embraced NFTs while museums, which normally require 3-8 years for exhibition programming, made acquisitions within months. It was perhaps one of the first times in recent history where artists and collectors didn’t patiently await institutional validation but instead forced the gatekeepers to adapt to their velocity. In many ways, NFTs represented the first real disruption to the art world’s carefully cultivated pace in generations.

Traversing Mediums

My transition to Sibylline Labs in January 2025 brought stark contrasts. Gone were the cautious deliberations and hierarchical approvals of the art world. Instead came an environment where experimentation wasn’t just tolerated but expected, where mistakes were reframed as data points, and where Mark Zuckerberg’s “move fast and break things” wasn’t just a slogan but an operational philosophy.

The move wasn’t seamless. The first few weeks brought polite nods from colleagues as I obsessed over pixel-perfect builds and exhaustive documentation. Finding balance required unlearning habits ingrained through years of proof-reading and white-glove service. What was most surprising wasn’t the technological learning curve but the cultural one; in the art world, expertise is demonstrated through encyclopaedic knowledge of the past; in tech, through the capacity to envision and create the future.

Yet beneath these differences lie unexpected commonalities. Both worlds deal fundamentally in human expression and value creation. Both require navigating the tension between inspiration and implementation, between vision and execution. The languages differ, but the underlying grammar remains remarkably consistent.

The Transferable Canvas

An art history background has proved surprisingly adaptable to working with these new tools, not despite its seeming irrelevance but because of it. Art history teaches pattern recognition across centuries and civilisations. It reveals how technological shifts catalyse cultural transformations. And crucially, it shows how new economic structures reshape creative expression, regardless of artists’ embrace or resistance.

The consequential flow—new technology spawns new industry generates new wealth fuels new art movements—has repeated throughout history, from Medici banking patents funding Renaissance masterpieces to Silicon Valley fortunes building contemporary collections. Understanding this pattern provides perspective on our current moment that pure technologists could easily miss.

It is often overlooked in technological discourse that artists have frequently been vanguards of technological innovation. They should not be seen as merely users of tools, but pioneers envisioning possibilities before technical infrastructure could fully realise them. Kevin McCoy created the first NFT in 2014, years before the technology would explode into mainstream awareness, essentially inventing a mechanism that is likely to become the standard for the value transfer of unique digital goods. And in 1973, Harold Cohen began developing AARON, an AI artist that created original works, decades before today’s generative models captured public imagination. This pattern reveals a profound truth: artists don’t merely respond to technological change, they often both anticipate and shape it.

Perhaps most importantly, art history provides a framework for understanding adoption curves and cultural resistance. Every revolutionary movement—from Impressionism to Abstract Expressionism—faced initial rejection before eventual canonisation. It is this historical perspective that tempers the frustration of explaining emerging technologies to sceptical audiences.

The New Renaissance

Technological developments in recent years have brought us to the precipice of a fundamental shift in creative production. AI agents aren’t merely a new tool but a new cognitive partner, one that will transform how we conceptualise and produce workflows and culture more broadly. The current dismissals (“where’s the human touch?”) echo critiques levelled at photography in the 1840s and digital art in the 1990s. As younger generations with more malleable perspectives become consumers of these tools from childhood, the questions of authenticity and ‘human touch’ that trouble today’s world could easily be fleeting discourse.

The current transformation extends beyond creative fields. The democratisation of technology creation—what could be called the “end of the SaaS moat”—puts unprecedented power in the hands of non-technical users. The tools available today allow anyone with creative vision to prototype and deploy sophisticated applications without writing a single line of code, a term known as ‘vibe coding’. The recent explosion of applications like Cursor, v0, and LoveableUI are great starting places for anyone new to the term.

We stand at a watershed moment comparable to the agricultural and industrial revolutions. The invention of the plough didn’t merely improve farming efficiency; it fundamentally restructured human societies, releasing vast populations from food production to pursue specialised crafts, scholarship, and art. Similarly, AI tools aren’t just improving productivity; they’re dismantling the technical barriers that have segregated creators from builders, visionaries from implementers, strategists from executors.

The Creative Destruction

Yet this technological bridge between domains isn’t without profound implications. As Joseph Schumpeter’s concept of “creative destruction” suggests, these technological shifts don’t merely optimise existing systems; they fundamentally restructure how capital, labour, and value flow through society. With these new tools, the cycles of creative destruction are accelerating at an unprecedented pace, which should raise critical questions about how our economic systems can adapt.

Consider SpryngTime—a free alternative to DocuSign built in just two days using ChatGPT, Cursor, and LoveableUI. Examples like SpryngTime illustrate the potential inbound collapse of traditional software development timelines and cost structures. The value traditionally captured by SaaS companies is rapidly evaporating as AI-powered development tools democratise creation. What we’re seeing is likely a fundamental shift in value chains: from software applications themselves to the underlying infrastructure that enables their rapid creation.

This redistribution of value appears to be flowing in multiple directions simultaneously. On one hand, to the foundation models that power these tools and the massive compute resources they require. On the other, to individual creators and studios who can now build and deploy sophisticated solutions without traditional development resources. The middle layer—the conventional software company—faces unprecedented compression.

As such, traditional venture capital models appear to be entering uncharted territory. When startups could once raise millions to build software products over multi-year development cycles with large teams, they now confront a reality where individuals using AI tools can create competing solutions in days at virtually no cost. This raises profound questions for VCs: Where do you deploy capital when the software itself is no longer the scarce resource? Does investment shift upstream to foundational AI infrastructure and cloud hosting providers like AWS, or downstream to distribution and brand? When anyone can ‘vibe code’ free alternatives to established SaaS products over a weekend, the traditional software investment thesis becomes increasingly difficult to maintain.

We’re potentially witnessing the early stages of an economic paradigm where value settles in new concentrations: the AI infrastructure providers at the base layer, the platforms that distribute and monetise at the top layer (those like Cursor), and a new class of augmented individual creators and boutique studios who can operate at scales previously requiring large companies.

These changes could represent the most significant restructuring of economic value since the industrial revolution. The parallels to manufacturing are striking; just as mechanisation first democratised production before ultimately leading to new forms of industrial concentration, AI tools are creating a momentary democratisation that may eventually consolidate into new power structures. The question isn’t whether power will concentrate—it’s where and how it will concentrate as these technologies mature.

The Hybrid Future

What’s currently unfolding isn’t just about improvements in model performance or efficiency, but the emergence of AI agents capable of sustained, goal-oriented work. These systems don’t merely augment specific tasks but can independently navigate complex workflows: researching, synthesising, creating, and implementing with minimal human intervention. What began as tools for discrete tasks are evolving into collaborators with increasingly sophisticated understanding of context and intention.

This evolution demands a corresponding shift in how we position ourselves professionally. As AI capabilities advance, a shifting landscape of professional value is emerging. While deep technical specialists remain essential, particularly in AI research and infrastructure development, a new category of contributor is rising in importance: the deep generalist. These professionals combine broad knowledge across multiple domains with the ability to synthesise insights and articulate vision. They serve as crucial bridges between technical capabilities and human needs.

Understanding cultural patterns and human motivation provides the contextual foundation that connects technical innovation to meaningful human impact. In a world where AI can increasingly handle specialised tasks, the ability to navigate across disciplines and see unexpected connections becomes uniquely valuable. The old boundaries between disciplines are dissolving, replaced by fluid interconnections of knowledge augmented by increasingly capable AI systems.

For tech, this lesson is crucial: without cultural and historical context, even our most advanced systems risk creating sophisticated tools disconnected from human needs and experiences. The “move fast” philosophy requires balancing with thoughtful consideration of cultural impact and ethical implications. This isn’t about slowing innovation but about enriching it with deeper understanding of the human experience it ultimately serves.

As these tools democratise access to technological creation, the emphasis shifts toward uniquely human capabilities, where traits like empathy, ethical judgment, cultural awareness, and creative vision are paramount. The most valuable skills become those that machines cannot easily replicate: the ability to understand human needs at a deep level, to navigate ambiguity, to build professional relationships and networks, to enrich culture, to make value judgments, and to communicate across different domains of knowledge.

As AI capabilities accelerate, the value of human perspective, historical understanding, and cultural context doesn’t diminish but rather increases exponentially. From auction house rituals to algorithmic workflows, my journey between these worlds has given me the perspective of this technological revolution not as a rupture, but as part of the continuous thread of human creativity finding new expression, with each innovation initially resisted, eventually embraced, and ultimately transformed by the very cultures they disrupt.

I’ve been building on some version of GPT for nearly 5 years, the last 2 with our portfolio at Sibylline Labs. The following is a retrospective of our experiences building with AI this year, and the assumptions and outlook we’ll be building from as we go into 2025.

The Frontier Labs

A purposeful design choice made early was to be model agnostic for each use case, selecting the most appropriate model for each based on a qualitative (vibes) and quantitative (cost benefit) analysis. This decision has shown outsized returns for us, especially as the gap between state of the art closed across the major players.

This is not an exhaustive list of the frontier labs and their efforts. Notably in recent weeks, the Chinese lab DeepSeek has released an incredibly impressive model, albeit one that we’ve no direct experience with, nor its predecessors, and thus won’t go into detail in here.

Anthropic

Anthropic will continue to win by innovating with builders on “just good enough tooling”, underpinned by clearly state-of-the-art models. I think the first signal from them here was optimizing around the use of XML tags as the underlying parsing mechanism for responses. These are demonstrably easier to parse out from complex and often inconsistent text responses than JSON. This was, at least to my knowledge, unprecedented at the time and has proven to be a trivial yet powerful interface for testing and integrating new prompts that might not have structured output consistency yet. If I remember correctly, there was even a paper demonstrating how forcing a JSON response from LLMs actually handicapped its reasoning capabilities, leading many teams (us included) to perform a 2-step LLM call to first execute the reasoning, and then format a final structure using a weaker model.

Google

Google will get its platform offering figured out, and with it will become the real “startup killer” that OpenAI isn’t turning out to be. RAG-as-a-Service, Code Execution, Internet Search, and Data Connectors (to GSuite at least here) are all nascent startups that are finding their positioning in the market that now have direct competition with Google. This is enabled by both their immature studio offering and the highly robust Google Cloud Platform, capable of serving not just their own demonstrably SOTA models but open source and partner models in a single, battle-tested platform, where these kinds of integration points are a config item, not a whole third-party integration.

Even now, I believe the market is underpricing just how robust and thought-through Google has made their vertical integration of the AI stack. They are capable of running SOTA models at a fraction of the cost of other labs and PaaS providers thanks to incredible custom silicon. From the offset with their first Gemini release, their “free tier” offering was so generous for the model’s performance that it was nonsensical, at least to us, to not deploy it for some use cases.

As Google reorients its execution capacity and ability to attract and retain outlier researchers around its re-found “mission”, even if it’s consistently lagging other labs on SOTA releases, its ability to deploy unit economics that no one else can come close to will force competition into making difficult decisions. It’s worth noting that many researchers, notably in the UK, highly value the stability and “established” offering of larger firms, rather than the scale-stage firms other labs offer.

OpenAI

OpenAI will continue to focus on its position as the “leader” in research of models themselves, rather than a holistic “AI platform” offering. Custom GPTs and other “side effort” retail-oriented developments will continue, but not as the core thesis. OpenAI is reporting significantly more usage on its retail ChatGPT offering than API usage compared with Anthropic, though I believe these numbers are speculative, not official. Accordingly, my thesis here is:

1. There is enough “ideas and research” being explored/backlogged and more than enough capital (and notably ability to raise future capital if needed) to chase them down to actually getting to a near AGI model.
2. ChatGPT is still a data asset class to them. This is reaffirmed by copious evidence that synthetic data works, and the introduction of “thinking” models as released with o1 where their LCOT can be used. We know of at least one open-source lab that is training their smaller models on the long chain-of-thought outputs from their reasoning models.
3. Deploying an operating fleet of weak/strong models to cost-and-performance-optimize faces as many, if not more, SRE challenges that they’re keen to battle-test. Their most recent outage, a well-known Kubernetes DNS-based issue, demonstrates that they’re still left on the maturity curve of SRE teams, and this provides a great testbed for them to refine that.

High-profile implementation deals with major firms will continue at pace. The Bay Area might understand tech, but as always struggled to understand the “irrationality” of all other major business sectors. High-profile tailored implementations for Magic Circle Law and Big-4 Professional Services firms, even if via a “GPT wrapper”, provide not only market positioning but insight to researchers about how the “AGI we already have” works via market forces.

It’s easy for AI labs to optimize for “frontier math” and coding benchmarks, but do they know how to design effective post-training for the intricacies of cross-border, cross-jurisdiction, cross-language, cross-precedent, subjective-outcome problems such as legal disputes between Germany and The UAE? This example is not random. It’s a requested use case we have been working on with a regional partner for 7 months now that has forced us to design and implement a myriad of interesting techniques to address a non-trivial list of edge cases and considerations to get even a base-level result.

Meta

Meta will obviously continue to double down on leading open-source SOTA models and the ecosystem benefits that both they and we receive from that, as seen in Zuck’s many public statements about why this is their strategy. In fact, I expect the “lean” on the OSS community to build around the models to only increase. I doubt they will bother designing and releasing a long-chain-of-thought framework or tooling for Llama 4 when within a week someone will have shipped a better one “for free”. In line with that, I wouldn’t be surprised if we see some major capital deployments from them out into the OSS community for projects they consider strategic or at least worth taking a bet on.

I also get the gut feeling we’ll see some incredible work in the small/local model space from Meta. A footnote found in their new strategy is an acknowledgment of the shifting terrain around them regarding data and privacy. Zuck even once hinted in passing about a federated social platform akin to BlueSky. Small local models would be essential to any strategy here, as would the models necessary to create a good UX for their re-entry to the hardware market via wearables.

The Vertical Platforms

It’s been a great year for platform players that had successfully positioned themselves for the unfolding landscape. We’ve found ourselves increasingly leaning on two of them as core strategic capabilities.

Vercel

Vercel has consistently proven to be a platform we can double down on throughout 2024. The team there is clearly very switched on and has a clear view of where and how they want to go. v0 has become one of the most essential tools of our workflow, and yet a rounding error on our total tooling bill. It being designed and built by other serious builders with builders in mind has meant its addition and establishment in our workflows was so natural it was almost a foregone conclusion.

Vercel’s new AI SDK offering alongside NextJS failed to deliver enough value from “yet another framework” for us for a long time, though demonstrably not for others. That said, it has rapidly matured into a powerful and production-ready SDK. This is a clear testament to the underlying culture within Vercel about how to build developer tools.

I fully expect Vercel to eat away at use cases both above and below them in their vertical with thoughtful, production-ready offerings delivered with their usual impressive velocity, which we shall be all too happy to consume.

Replit

Replit has clearly gone from strength to strength as its full end-to-end thesis continues to be realized. It has become my go-to response to clients and friends when they ask where to start trying to actually ship things. I downloaded Replit whilst on a night out in London to ship a moment of inspiration that came to me, a moment that really drove home its powerful value proposition of making good use of its well-executed AI Agent product.

It’s not quite at a Soften level of noob UX, but also not as flexible and powerful when desirable (which is fine as it’s clearly not the target user), but I don’t think it’s far away. Whereas Vercel started from the Developer Platform and moved vertically, Replit started with the best web-based IDE and moved vertically. My intuition is that this interface choice is now going to become constricting should they double down into it.

Their AI Assistant product is, albeit well-executed, just a chat AI over the repo. I think there are so many opportunities for exploration here, such as a thin abstraction layer over confusing directory structures of repos for new beginners. Perhaps a more “proactive” clippy-like experience that attempts to understand more ancillary actions, not just code generation. From personal experience, trivial low-hanging fruit they’ve not addressed is an agent that can actually perform Replit/Repo-level actions. I spent more than 30 minutes in disparate documentation and unintuitive settings trying to change a simple config line that could easily be an AI utility.

I look forward to putting more leverage into this tooling. As the barriers to entry and inertia for “technical work” continue to lower, our senior engineers and product people are able to move at a compounded velocity to those outcomes. Where v0 fails to provide a persistence layer, nor a reference backend implementation, Replit can, enabling executives and other customer-facing resources to independently create and iterate on concepts and sales assets without forcing context switching from engineers.

Developer Tools

One of the most immediately obvious set use cases for language models was around code generation and developer tooling. This reached an almost mimetic level this year as YC announced the backing of almost a dozen different “AI Code Copilots” or similar.

Devin

The impressive general availability launch of Devin led to one of the fastest purchases for the studio this year. As a purposely lean team, we use Slack as the control plane for orchestration and communication, which Devin appropriately aligns with as a first-class citizen. Where normally we would need to look to custom integrations to our issue tracking tooling, our ability to also connect that to Slack and just import that context in a simple manner feels like an incredibly natural addition to the team. I purposely chose “team” here, not “workflows” where normal new tooling intersects, as it truly introduces a whole new set of workflows and resourcing capabilities that would typically require a full-time hire.

Popular complaints about Devin are largely a result of their own marketing positioning it as a whole software developer asset, when pragmatically right now it’s a talented, book-smart junior developer, and thus needs to be treated as such. Many enterprise teams who’ve had this trust thrust upon them may vent their frustration at how it’s not the magical AI developer they were sold. As someone who is viscerally aware of how much it costs (I pay the damn invoice) and what we get for it, Devin is a fantastic product that I eagerly await future versions of.

Cursor

Public sentiment of Cursor is fascinating to watch. I don’t think I’ve seen a middle-ground take that says “eh I’ll stick with it and see”. People either instantly fall in love with it or outright reject it. Those in love are the least interesting of the two. It’s built upon a good experience of a well-executed product, building on the popular and powerful open-source software that is VSCode. They clearly have at least some kind of insight into effective AI co-piloting.

It’s easy to dismiss those who hate Cursor as dinosaurs that hate it because they hate all AI code assistance. I put zero stock in that thesis. My intuition here is that these are the same people who also objectively do not enjoy “dynamic pair programming”, that is to say, having someone (or something) modifying and changing the environment that you are working in. They’ve built AI assistance flows with an independent mechanism (chat or otherwise) where they can control the flow of change. I am one of those people. Originally starting with just Anthropic (both console and chat) and then progressively leaning on aider, what in my opinion is the perfect interface for a code assistant.

I wonder how penalizing Cursor’s choice to exclusively build around a standalone VSCode fork, rather than a cross-platform plugin, will impact their longevity as the 20 other YC-backed co-pilot startups chase them down. Perhaps there is a market for N number of winners here, but as touched on vis-a-vis Replit, as the interface gradually moves away from the editor, what does the future for them look like?

Closing

We’re still incredibly early. Across foundation models, infrastructure, tooling, and product design. Despite the velocity of releases and shifting landscape, we don’t expect a slow down any time soon. Our thesis is that vertically integrated, AI-native apps, with newly enabled interfaces and experiences present the most exciting opportunities, and thus where we’ll be spending most of our time.

Background

Tornado Cash is what we call a “mixer”. A special type of privacy-preserving tool that runs on permissionless, decentralised blockchains. That’s a lot of words to the uninitiated, so let’s break them down.

A blockchain is an immutable ledger (read: database) of who sent what tokens to who, and when. Some fancy cryptography ensures that this ledger cannot be tampered with, and thus is always valid. This is the power and simplicity of a blockchain. This fundamental design means that we can always trace where money or on-chain assets originated from, and who they’re going to, a sort of directed graph that shows us where money started and every step it took to get to the other end. A mixer is a simple smart contract (read: program) that runs on these chains that accepts deposits from anyone, and “mixes” it all together in a way that makes whoever redeems a deposit impossible to distinguish from the depositor themselves by breaking the observable link between a deposit and a withdrawal address.It proves – very simply – a mechanism for creating on-chain privacy for transactions.

Permissionless is a term we use to describe the fact that no-one needs to grant permission or access to these blockchains or tools. Literally anyone can create an account on these chains and start using both them, and the smart contracts on them.

Decentralised is a term you’ll hear thrown around an ever increasing amount, though with each actor using it differently to suit their needs and interpretations. We’ll be describing it simply as “neither governed nor controlled by a single defined actor”.

So Tornado Cash was, put quite simply, a tool that anyone could use to preserve the privacy of their on-chain transactions.

The Foundations of the West

Like so many of our peers, we are firm believers that the principles and ideas of Western liberalism are the correct ones.

Freedom, your basic rights as an individual to speak, transact, and live without fear of unjust persecution, arbitrary interference, or subject to inspection by some authoritarian body.

We have written previously here about how we believe your digital identity and your financial one – are in our modern age – one and the same. As the frontiers of technology continue to move forwards, particularly in the realms of financial payments, settlement, and digital identities, this truth will only become more apparent.

Since the dawn of democracy, it has been made abundantly clear that a healthy and functional courts system is essential in ensuring these rights are upheld and not abused by those we elect to positions of authority. As the markets that formed in these democracies evolved, so too did the essential role of the regulator to ensure consumer protection and fair play in the market.

It should come as a stark warning then, that one of these democracies has gone so far as to pass a verdict of guilt for a crime that has not been committed, but for one that may be committed in the future.

The Dutch court was of the opinion that it should have been ‘foreseeable’ to the software developers when they were writing the Tornado Cash open source code that, at some unspecified date in the future, money originating from crime could be deposited into the smart contracts, taking advantage of the concealing effect of the tool. It was on this basis of “foreseeability” - that these immutable contracts might one day be used by a sanctioned entity to launder money - that Alexey was found guilty.

The court viewed Tornado’s privacy protection function as a function to conceal - which amounted to conspiracy to money laundering under Dutch law. By doing so, the judges assumed there was no legitimate use case for privacy and treated the developers as responsible for the creation of a tool of potential misuse.

One doesn’t need to be a Dutch legal practitioner to realise the profound misapplication and contortion of the law in order to secure a conviction.

In a case of this nature under English law, in order to prove conspiracy to commit money laundering, the prosecution must prove that the defendant: (1) agreed to conduct, or to attempt to conduct, a financial transaction; (2) knew (suspicion is not enough as a state of mind) that the property involved in the transaction represents the proceeds of crime, and (3) proof of intent to conceal the proceeds.

None of these elements exist in the case of software developers who create and deploy immutable smart contract protocols, and none of them were proven in this case.

Never in the history of liberalism has anyone ever been found guilty of a crime that could potentially be committed by another, third party, at some time in the future. A truly terrifying precedent to be set by the court. This conviction grants a government unlimited power to prosecute any software developer who writes code that is later used by a third party for nefarious purposes, merely because the developer becomes aware of that later use. With no limiting principle in place, nearly all developers who create open-source software would be exposed to criminal liability for activity outside of their control years or decades later. Put simply, the Dutch court’s theory of liability has the chilling effect of rejecting core principles of due process and the rule of law.

Building for new Precedents

With the precedent now set that you can be imprisoned for writing and releasing code that could be used by a sanctioned entity, all the builders of the world should be taking heed of the warning.

Whilst the current target – privacy preserving tooling – may fall outside the scope of what you are working on, given the gravitas of the ruling, we don’t expect the contagion to stay there. Anything from DeFi apps to SocialFi that “may” be used by sanctioned entities is now on the table for the court, and you should consider yourself at risk.

With all that being said, we do believe there is sufficient clarity for a robust direction forwards. This essay will outline some of this thinking, though it is not to be treated as any official kind of counsel, or legal advice. If you do feel in-need of official guidance, please don’t hesitate to reach out to us.

The basis of our thinking builds upon a handful of main factors;

• That the regulatory outlines and requirements for compliance are increasingly clear;
• In our (widely shared) opinion, the correct way to build on-chain is permissionless and immutable;
• That privacy is an important individual right that needs protection; and
• That there is clear precedent set for this.

Regulatory Clarity

Albeit far from where we’d really like to be in terms of absolute clarity – as expected in maturing market sectors – we are slowly arriving at clarity on the regulatory and compliance requirements for crypto and virtual asset providers. For the builders in some jurisdictions, for the better. In others, the worse.

The requirements for sanctions and AML screening, the restrictions and controls necessary to be in-place to achieve minimum standards for certain types of digital assets, and general diligence necessary to be taken seriously, are broadly speaking becoming defined. Regardless of your personal ideology and views of the sector, this is a good thing.

The irony of writing these paragraphs given the ruling we’re discussing as the centre-point of this essay is not lost on us; simultaneously the ruling itself sets a degree of clarity. It’s now clear that any privacy preserving software, on-chain or not, is now within scope for legal persecution on the European continent. Risk models should be updated accordingly.

Technical Precedent

It is true that no smart contract deployed to a blockchain can be modified. Once deployed the contract itself is immutable. That being said, it is in fact possible to change the behaviour of a protocol through various technical mechanisms that are outside the scope of this essay. just take our word for it (lookup upgradable contracts if you’re really that interested)

It has been an open debate for some time as to whether this approach should be endorsed – particularly for certain types of protocols – as the right way forwards.

The benefits are obvious: the ability to upgrade and improve the behaviour of a protocol with what would be a seamless experience for users. The contracts are “upgraded” (technical implementation notwithstanding) and the users benefit.

If smart contracts can be “upgraded” to improve the protocol, it stands to reason that they can be changed to enact behaviour to the owners’ own benefit only – a compromise of the protocol if you will. The idea that protocols should be immutable on-chain - i.e. being able to be trustedabsolutely such that we can read the code,in-perpetuity, in the knowledge they cannot be changed - is compelling. Uniswap as a protocol is a clear and established demonstration of this.

The Uniswap Precedent

There are now 4 versions of the Uniswap protocol available. Each one a wholly new deployment of smart contracts with no association to the previous. This was by design. The contracts could not be “upgraded” to automatically distribute these new features or ideas to users. Each user of the protocol – both sides of the transaction – had to select which version(s) of the protocol they wanted to use. To this day, there are many newly released tokens that are distributed on older versions of Uniswap for various reasons.

What this does mean however is that as time moves forwards, the trust we can place in the contracts to always behave in that manner, and not have security risks increases. Indeed, if there was a critical flaw in the protocol that a threat actor could exploit, it almost certainly would have by now.

It also provides the creators of Uniswap with a very powerful defence against the rising tide of regulation and scrutiny. Uniswap have reasonable neutrality as simply creators of “lower level” parts of the system.

It’s that “lower level” aspect that is quite critical. Even the most technically literate and crypto-native of readers of this essay have likely never actually interacted with Uniswap directly at the protocol level. They’ll have almost certainly used the Uniswap UI – the officially endorsed product implementation – to perform whatever action they were trying to.

Layer 8 Problems

There is a model we use in computer science called the “OSI Model”; a dated, albeit effective tool for understanding how our interconnected digital world works, each “layer” in the model a different position in the card stack of abstractions we use to connect systems, one on top of another until we arrive at what you experience as the modern internet.

A better view of what this means in practice would be the above – an evergreen xkcd comic – about the handful of developers who (thanklessly) continue to develop and support critical “low level” packages and libraries we use in our daily lives.

There’s that “low level” denomination again.

If Alexey is guilty of writing code that “foreseeably” could be used by sanctioned entities, then so are the developers of the encryption, networking, and browser libraries we all – even more so the government – use daily to protect our digital activities; from not just authoritarian inspection, but from the general malice that permeates the world.

The courts would never come after developers of anything lower down in the OSI stack; a model that we feel that today, does no longer accurately encapsulate the evolved complexity and abstractions that come with blockchains. In any other “traditional” system, implementing the relevant controls and regulations at the top of the stack was deemed sufficient. We don’t feel that there is any substantial reason for this to change.

The design and development of immutable, composable protocols – financial or otherwise – specifically where they are to be used on decentralised, permissionless blockchains, is no different from developing any other “low level” protocol. These protocols must be immutable for any credible level of trust to be placed in them, and thus are impossible to comply with evolving regulatory requirements.

Assuming the same set of developers who would likely go on to build a high-level, user facing application that simply leverages this and potentially other protocols, go on to implement any and all pertinent regulatory requirements and controls, there is no reason to believe that these developers had any intention of sanctions busting or non-compliance.

Close

We should be thankful that there will be a second court case, that for Tornado Cash co-author Roman Storm, held in the US; due instead this time for a jury trial, and not a Dutch three-judge-court. We shall be keeping a close eye on how that unfolds, what evidence the US prosecutor can actually bring to bear to prove the charges, and how the jury interprets each position. It is our deep personal hope that sanity prevails, and that the US maintains its foundational commitment to individual liberties, in a time when its democracy remains again tested.

We live in truly unprecedented times; where the idea of something that isn’t a person can be subject to sanctions, and that a person can be found guilty of something that may or may not happen in the future. We champion and support anyone standing against the alarming rise of illiberalism wherever it can be found, and look forwards to a future where personal and financial freedom for everyone is attainable in any format, private, or otherwise.

Sibylline has also been active building AI products alongside our customers, and in doing so we’ve started to form our core thesis on how we think good AI products are developed, both today as you try and move as quickly as you can, and in the future as the ecosystem and tooling matures.

The high level of our thinking is centered around the idea of “agentic behaviour” or this idea that AI applications should start to show emergent properties of doing things without necessarily needing to code or program it to know how to do the thing. They will begin to understand ancillary requirements and behaviour without the designers and developers having to force it into a straight jacket.

Foundational models will continue to get increasingly powerful, and the locally available compute available to run them will scale with them, but there is general consensus that the future is not “one all powerful model” but rather a series or collection of models; each tailored to their respective domains and roles.

This concept of “domain specific models” echoes similar thinking in systems engineering we’ve built upon over the last 2 decades; the Domain Specific Language. It also makes sense from an anthropological and neuroscientific view – areas still, in our view, heavily underutilised in the field of AI applications – as to how knowledge and intelligence is structured and applied more generally in society.

Models

We’re firm believers in remaining “model agnostic”; the idea that we’re not going all-in on a single foundational model or provider and attempting to make it work for every use case. It’s widely accepted that at the current state of affairs, OpenAI’s GPT-4 is the most powerful model in the raw sense of the word. This however does not paint the full picture.

GPT-4’s power comes at a cost, one that often isn’t necessary for a lot of use cases. It’s also limited – as with all models – by it’s “context window”, or how much information you can provide it in a single session. Conversely the Claude-2 model by Anthropic is capable of holding more than 3 times the context window, and can be much more innately suited to tasks that require handling that many tokens.

Fine tuning models for specific use cases has demonstrated a clear track record of being able to outperform larger ones. We expect this trend to continue. In doing so it presents a compelling case that data will provide one of the strongest moats to AI model and product development. Organisations with mature data operations would benefit from now applying the leverage possible from fine-tuning their own models to tackle internal inefficiencies or broader product opportunities.

Interface

A large majority of the initial AI products we saw on the market were chat-based products. Inspired by the ChatGPT product that captivated the entire world the possibility of these models, lots of people sought capture their own slice of the hype-run with their own niche takes on the “AI Chat” product, with varying degrees of success.

Chat

We ourselves experimented with this format with our Delphi product; a cybersecurity domain specific model trained on a cleaned dataset our engineers have been crafting for 3 years. The chat app was “nice” and provided a level of utility otherwise unavailable on other services, especially on its own domain, but itself wasn’t compelling enough to move people away from instinctive tooling such as the normal ChatGPT. A phenomena most of these apps went on to experience.

The model, infrastructure, and engineering – particularly around RAG – proved invaluable in the development of the products that came after it, both internal and for clients.

Any chat-based interface should be predicated as an embedded feature within an app that users are already using as their set of core daily apps;

• Data intensive application such as dashboards can leverage this with a “what am I looking at” style feature. We’ve had great conversations with Elastic and how they’re thinking about this application in their Kibana product.
• Core workspace platforms are already well on their way here to embed a “lets help you write” feature within their products; Notion being a great first mover in that space
• User support products have routinely turned to AI to enhance their products; even more so short-lived are the days of static support flows. Ask any question about a product or service and get answers direct from a well trained model.
  • FAQ’s are not long behind in that regard.

This list almost certainly isn’t exhaustive, but provides a sense of how we think about applications of chat interfaces. We have another essay due this month to watch out for on “Chat is not the interface”.

Reasoning

The open debate around whether these models are truly “reasoning” or “remembering” is far from concluded. I’ll save our full thoughts for another essay, but at a high level we do believe that these models are capable of a form of reasoning. Whilst we recognise just the sheer volume of information these models are trained on, the “it’s just remembering” side of the argument lacks a certain understanding of how human intelligence itself reasons about things; see our earlier point about the lack of applied neuroscience to the space.

From our experiments and experience building battle-tested applications with these models, we’re convinced of a base form of reasoning, and it’s here that we’re excited about exploring applications of AI.

Understanding, contextualizing, and then presenting thoughts over large amounts of information is still one of the most underutilized applications of AI models. It’s for this reason that a lot of people suspect that these models will replace knowledge workers such as consultants. These people do not understand why consultants are paid what they are.

We’re even more excited about problem spaces that were largely left unnoticed by traditional consulting and was left to more niche or boutique players. Paperwork intensive spaces we’re also bucketing in this area and actively exploring with clients. Incoterms, planning permission at both a retail and corporate level, litigation, and compliance reporting. Areas that have been left largely undisrupted by technology in terms of how they operate (no, SaaS did not disrupt the Magic Circle)

Agents

As we touched on briefly at the start of this essay, we believe that the most robust AI products will trend towards AI Agents or as the base case, exhibit agentic behaviour.

Agents are largely defined as system that is designed to perceive its environment and take actions to achieve a specific goal or set of goals. They don’t need to be specifically programmed or told how to achieve the desired outcomes or goals, but can make assessments and reason about decisions of their own accord on how to reach them.

The thinking behind products trending towards agents is that, well, a lot of the best ones already are. The expectation of AI driven products is that as the models become increasingly more powerful, the “magic” factor that customers will expect from them will increase in-step. This “magic” factor is most likely going to be derived from agentic behaviour.

This also makes sense from a product perspective. A well designed and well equipped agent should be easy to build upon; providing it new capabilities and therefore features should be relatively simply, as opposed to hamstringing basic AI features into a single product vertical or segment.

Modern AI models are capable of far more than just basic completion and conversational behaviour as well. Pioneered in open source development, but popularised (again) by OpenAI, the use of “functions” or the ability for models to call and invoke other services has rapidly matured in recent months.

These were initially just used to allowing the model to access up-to-date information about its surroundings, the classic use case shown by OpenAI in their examples is getting weather information from an API. This is powerful in its own right, but the true power came unlocked when they were applied to writing data. Tasks such as updating records, or both chained together into more complex state changes such as collecting information from multiple systems, reasoning about it, and then writing an new update (how much longer systems like Jira will still look like their current form is an interesting thought experiment here)

Systems Thinking

It’s right around here that we think a lot of agent builders could benefit from the learnings of years of systems engineering.

Whether it comes from a user, an external service, or the result of an LLM, final completion or function call regardless; event driven architectures have been proven to be one of the most effective and scalable systems in modern computing.

A lot of the earliest “AutoGPTs” – agents designed to manage their own agents – ran into all sorts of predictable state management, visibility, and orchestration problems, that could have been easily avoided with the introduction of basic event driven architecture patterns that we’ve used in industry robustly now for years.

Simple concepts such as topic queues and routing, when coupled with the agent’s ability to read the data it needs, can be combined to create a multitude of agents, each with their own domain specific tasking, where each can be prompt and tuned to their optimal performance, and can be coupled with years worth of mature tooling around distributed tracing and insight into complex systems.

We’ve played around with some of these concepts in some of our most complex AI applications to great success. I suspect we’re only just scratching the surface here on how these engineering concepts can be applied effectively to agent design, but do expect a follow up that goes a little more deeply into this particular space as we discover more.

Epilogue

We’re all still exceptionally early. I’ve been a professional engineer for approaching a decade now and I’ve never seen a single vertical move as fast as AI has in the last 18 months, it’s fascinating.

The hope is that some of the thinking here might benefit other thinkers, experimenters, and builders in the space. We continue to build and learn more about the capabilities daily, and will continue to share what we learn along the way.

We’re confident that, given present understanding and experience, that this thesis should stand up. We’ll review it in a years time to retro what we got right, wrong, and weird; updating it with another years learnings along the way.

Just over a month ago we launched Delphi. A demonstrator chat interface we put together to answer questions about cyber security for our clients. Anything from “is this a threat” to “what does this compliance control” mean.

That said, we were never convinced that chat was the optimal interface for working with LLMs. Our product lead was particularly enthused about finding the “real” way we should be bringing the power of these models to products.

It must have been serendipity then that a few weeks after launch, the 2 main sources of inspiration landed before us in very quick succession. First, one of our early power users asked us “hey, I know I can upload files, but can I give it an email? I think this one is dodgy”. Followed quickly in rapid succession of receiving a screenshot of an email from my mother with her asking “is this thing safe” (a habit she’s been a fan of since I started my career in cyber security)

It was made clearly evident then, that one of the most valuable use cases we had was looking at emails, and incidentally this meant that we had a solution to our interface problem at the same time.

Email, broadly speaking, is a “solved problem”. There are countless, very good tools out there that enterprises can use to tackle it. They’re not perfect, but when used in conjunction with other controls, you can have Pretty Good Confidence™️ that your estate is secured from these kinds of threats.

The challenge with these tools is that you need a certain type skills in your organisation to configure and use them. Most small organisations don’t have these kinds of skills in-house. Your tiny mom and dad shop that has been around for a few decades is likely still running the same tiny IT estate they have for years, and almost certainly hasn’t turned on all the fancy new features they have come out, because why would they, they’re not cyber security experts.

Users don’t want to need to install new software or tools, nor configure email plugins. They want to feel confident and safe to use email, but will churn quickly if frustrations arrive at getting there. Safemail removes all of those problems, instead moving the solution to the email button they already know and love. The forward button.

No configuration needed. No setup (literally, you don’t even need an account to get started). Just forward us any email, and within a couple of minutes you’ll get a reply from us telling you whether it’s safe or not, with some advice and next steps for in case you’ve clicked on anything.

Naturally the first thing done when launched was to get it Mom Certified™️ to ensure that problem was solved, and we’re happy to report that is has been.

Safemail is still very early. We’re big fans of shipping quickly and iterating so we’ve got some nice enhancements on the roadmap for the next few weeks, but we’re already happy with the results. One of our B2B SaaS clients is already asking if it can check every email against it since that’s where they do a lot of their commercials, an offering we’ll launch publicly in a few weeks.

Try it out for yourself, forward any email to agent@inbound.safemail.ai and get confidence in email back.

Cyber security is actually pretty hard. It’s even harder to get it “right”.

We live in one of the most powerful and enabled generations of all of history. The birth of the internet era has enabled literally anyone with a burning enough desire to solve problems at a scale previously thought impossible.

That said building digital products and services isn’t a complete walk in the park, and the internet can be a dangerous place if not treated with the respect it needs. The rise of cyber crime has led to market size of $7 Trillion by some estimates, larger than some of the worlds biggest economies.

Building digital products is great, doing it securely is

We know that not everyone is a cyber security expert, we also know that a lot of the terms and language is incredibly nuanced and vague, even confusing the experts themselves.

Getting answers to questions such as am I doing this right, what the hell is “compliance” and that alert looks scary what should I do? is never as easy as we’d like it to be.

This is where Delphi comes in.

The What

Delphi builds on top of our work with other AI models, specifically Large Language Models.

Delphi itself is a fine-tuned LLM using a custom dataset of cyber security information, alerts, and reporting that we’ve been compiling for several years now.

We’ve then built a very simple chat wrapper around the produced model that you can ask questions to, very similar to how other conversational AIs work, but specifically designed to help you get answers and next-step actions for cyber questions or problems you’re facing.

Today, it’s just a simple chat conversation. We started with just this as we wanted to get a feel and understanding for the kinds of questions people have about cyber and how they would ask them.

The upcoming future roadmap however, is incredibly feature rich, including;

• Directly plugging into your current cyber security tooling to automatically collect, explain, and advise about alerts or notifications
• Allowing you to upload files or links that you might have opened but suspect might be malicious, and get instant advice and feedback about it
• Pasting whole rows of vulnerabilities and having it understand and prioritise which ones you should start with.

Sounds good enough? Try it out here

The How

As we mentioned earlier, Delphi is an AI model specifically tuned to understand cyber security better, but it doesn’t stop there.

Whenever a question is given to Delphi, it first performs a lookup of relevant and contextual information that can better inform its decision making from both our massive and constantly evolving dataset of cyber information, and the contextual understanding of who you are and what you’ve previously asked it the more you use it.

All of this is then fed into a chat completion system that has been pre-prompted with the relevant context and requirements. We’re continuously evolving what these prompts look like to ensure we’re getting the best results possible.

Our product thesis for this being that the chat agent needs to both;

• Perfectly tailor the response to the user that is currently asking the questions, be it a cyber security professional, or someones grandmother that has unknowingly opened a file they shouldn’t have
• Provides concrete next steps that you can action with confident authority.

That last part is a lot harder. Getting language modals to be commital about their answers after performing sufficient reasoning about them can be hard, but we’re confident in our approach

In any case, we’re excited to see what people ask Delphi and see how it continues to evolve as a useful tool for both professionals and not over time.

It might seem almost cliché to be jumping on the AI bandwagon like so many others, but our partners have been building and shipping AI products back when good GPT models were in their infancy, and more than a year before the introduction of the ChatGPT product that popularised them.

Introduction

When we talk about the most recent advances in AI, they come across in 2 key main verticals;

• Diffusion Models - such as DALL-E and Mid Journey
• Large Language Models (LLMs) - Models such as GPT-4 and LLAMA

Whilst there have been incredible advances in the generative AI section, especially across theoretical sectors such as healthcare and neuroscience, our work is mostly focussed around the use and deployment of LLMs and natural language, and and is where this series will keep its focus.

But what exactly are LLMs and natural language? This passage from a soon to be released white-paper we authored for a client explains;

Natural Language

Natural language refers to the ways humans communicate using either spoken or written words and naturally arises in human societies as a means of communication. Most of the languages we speak on this planet fall under the umbrella of “natural language”.

Natural language is both complex and flexible, and allows us humans to express a wide range of ideas, emotions, and concepts, but it’s also innately ambiguous. It’s this ambiguity that computers struggle to understand, and where the field of natural language processing comes in. There are no fixed rules in natural language, and even if there were, a significant portion of the speaking population would not follow them; this is the origin of slang and local dialects.

Natural language processing instead turned to the world of statistics to glean insight into language and how to make it machine usable, with the most recent breakthrough in this field coming in the form of “Large Language Models”

Large Language Model

Large Language Models (LLMs) are a form of artificial intelligence built on top of deep learning architectures, initially such as Neural Networks, but more recently leveraging what are called “Transformer” models, an approach initially pioneered by Google

LLMs were popularised by OpenAI with their release of the ChatGPT product; a simple interface for conversing with their latest GPT3.5 LLM model. This model surpassed all models at the time at the tests available, and captured the imagination of the world as it demonstrated the real power of LLMs in action, producing some incredible works of writing, code, and insight with what seemed like trivial effort

In practice, this model was simply the industrial application of natural language processing. The contextual meaning of sentences being passed to it, and the expected reply to them, was understood and created through the intensive statistical analysis of a massive dataset on machine learning models with billions of parameters.

Natural Translators

LLMs are exceptionally good at getting from one form of language to another, be it from one interpretation of a text to another, between whole other languages, or where some of the most important advances are being made, getting from language that people understand, to one that machines do.

The process of getting from what users want to do in software to the machines actually doing it has been one of the most important and studied fields in modern software engineering. It drives everything from the foundational APIs and primitives we build upon, to the user interfaces and design choices that are made.

LLMs in this regard, are the translators we’ve always been looking for to help get us from the language of expressiveness and nuance that we humans use every day, to the language of rules that is necessary for computers to function.

There’s no clearer example of this than Github Copilot, a product designed to support engineers in writing code — the language of rules that machines are governed by — from simple human readable comments.

Prompt Machines

The use of LLMs is not however, just limited to translation. These models similarly fall under the “generative AI” banner for a reason; when given a limited input, prompt, or question, you can leverage them to generate or complete the content or answer for you. This enables them to fill interesting roles in the search and chat functions.

There’s a common misconception around models such as these that they’re sat atop some massive database of all known knowledge to answer questions, much in the same way a normal search engine is. This isn’t quite accurate (albeit yes the new Bing and Bard products to leverage AI with their traditional search engine databases). These models are trained on massive datasets that allow them to predict the answer – with incredible accuracy – to your question.

Sibylline was named after the Sibylline books, ancient books purchased from a Sibyl by the last king of Rome, and consulted upon during times of great crisis. The word Sibyl is derived from the ancient greek sibylla or prohpetess; those who would give answers towards those who provide good questions.

LLMs operate on a similar premise. The responses that these models create are formed from not just the input question that you provide, but also from the surrounding context, additional input information, and “prompting”. All of these collectively go in to draw the attention of the model to the next correct token, leading to the most accurate answer.

Or as Stephen Wolfram succinctly put it

Closing

The AI industry is advancing at a rapid pace, with innovations being unveiled so quickly it seems we might need our own platform to keep pace—an idea proposed internally and echoed by our clients.

As mentioned, this is the first in a series of posts about our work and experiments in the AI domain and how we guide clients in harnessing its potential. Throughout this series, we’ll delve into:

• Our strategy for identifying and validating effective AI use-cases and deployments to address real client problems, and avoiding hype-driven investments
• Technical breakdowns of how we build and deploy cost-effective LLM solutions for our internal tools and projects
• Stories from our journeys with clients, highlighting the process of unlocking the power of AI for their businesses

Look forwards to having you follow us for the ride!

If this were a different kind of coin less linked to USD, I’d be more charitable with their avoidant behaviour. But one of the key USP’s of any pegged stable is the trust in its value holding steady - the clue is in the name. The problem we now face is that Tether is a victim of its own success - it’s responsible for some 64% of all stable throughput, and when your only significant and repeating profit opportunities are arbitraging on unusual FOREX events, market cap becomes a very different question to reserves.

Much of the following discussion often begs the predicating question though:

Do We Want Tether et. al. At Full Reserve?

Tether might be caught in a regretful bind on this specific point, having committed themselves to it under a humbler vision. It might seem intuitive to assume that a provider of a tokenised national currency should always be at full reserve, the same way you assume that about your retail bank. Remember what happened the last time everyone assumed this about the banks? I don’t say this to catastrophise, but merely to point out that intuitions are often only good for bracket shooting in this space, especially when they’re predicated on old-order models.

I don’t see an inherent reason why a firm offering a token on a national currency necessarily must operate at full reserve, let alone a truly separate currency, provided partial reserve status was explicitly communicated in both cases. If we’re setting up a faithfully Hayekian currency market, then those with greater risk appetite, or appetite for whatever a partial reservist can specialise in, should be free to select those with partial reserve. Dare I go one further, much of the catastrophising around partial reserve re-entering the space is a 2008 hangover unjustified on the facts. Smart Contracts and Blockchains are beautiful technologies in this regard - no matter how strong your incentive is to game the system, as long as the amount of each currency is in circulation is public knowledge, errors correct.

Of course, the above doesn’t apply to Tether for as long as they’re not willing to audit and/or put some proof on chain. But one can imagine more complex systems of scaling reserves dependent on input and price fluctuations elsewhere. It’s doubtful how popular some might be - locking some inputted funds away for investment is anathema to a coin based purely on exchange. Simple interest rates might be a better mechanism, but then whoever’s offering the partial reserves needs to beat the market by an extra x%. Beyond this point my ideation is limited, but I think that’s enough to challenge any mechanical assumptions you might have made.

Movement Away From Tether

So what would be the effective difference between what I’m ideating on and Tether Gold? It grants tokenised ownership in gold, mixing confidence and convenience, and, accepting some rounding errors, its reserve seems 1:1 backed. Well first of all, a new project would be separate from any concerns about taint by association; Tether Gold lacks the market cap to hold up Tether if the latter falls. Second, gold is reliable, but has been permuted throughout history and the price is subject to short-run inflation as confidence in fiat continues to fall (correctly) - this would also apply to a successful new asset, but only after initial private sales and assignment. Finally, a new commodity underlying stables offers an entirely new paradigm, insulated and separate from any central bank who might want to switch back to gold in restoring confidence; purchases are already up.

But what kind of new commodity to switch to? It’s far from obvious, but ideal ground for research ideation. There’s the usual financial criteria, stability, abundance, portability, inert, so on. But, due largely I admit to cultural instantiation, any competitor must overwhelm the instinctive preference towards gold which the supermajority will feel.

The overwhelming concern which I’ve osmosed from true specialists in stables is fungibility and matching liquidity. The challenge occupying any Tether competitor will be resolving exactly the pain point they are under scrutiny for, without having a 64% market share to work from. How would a competitor go about establishing the sort of confidence necessary to challenge that inertia? My instinct is to run straight for institutional acceptance and show an overwhelming amount of supply in a commodity which can be bet on as consistently abundant. Demand can flex from there, inside limits. The external demand for the commodity in question should therefore be inelastic and its industries insulated from shocks on both sides. If subsidised or state-monopolised, though it pains me to say it, so much better for the issuer’s immediate acceptance and launch velocity.

Inevitable Fractionation?

Still, $83bn is a massive target to aim for. I’m doubtful that any new supplier of a stable could match it coming in cold - they’d either need a massively successful raise, a titanic private backer, or some massive capital pool aiming more for diversification and persistence into the future. Even better if you can combine these; but there will be a few out there who can fit that definition. More immediately, if the legal situation around Tether deteriorates further, questions about the future of the marketplace will follow if that 64% share can’t be quickly disseminated or retained with newfound confidence.

It then follows that a more competitive marketplace, with multiple successors sharing pieces of the pie, might be better for all involved. It will be easier for each to demonstrate whatever % of matched reserves they offer, and competition will place them under greater pressure to actually do so (the lack of this has insulated Tether until the law came knocking). The consumer gets a truer Hayekian private currency market, suppliers are incentivised to innovate faster not just on the technology underlying stables, but on making them understandable to the consumer. It’s easy for the technically oriented to underestimate the potency of adoption, but I can think of nothing better to draw attention, talent, and capital to the stables space.

So, what are the most immediate competitors? There are plenty of other stables offering Dollar 1:1 pegging, if not a 1:1 reserve. For the average consumer, I’d say that TrueUSD and USDCoin are straightforward betters. They are regularly audited to prove that they maintain full reserve for each token issued, though this raising interesting competition questions with CBDC’s. As and when the SEC gets its act together, or is compelled to by somebody getting a hand on the three-letter agencies, a digital dollar becomes a concern. And when the average American can straightforwardly deposit with The Fed, I struggle to see a niche for these Coins to gain new customers. Those already in, either for ideology or fashion, may maintain their position, but if the USP of these firms is a combined confidence in the dollar and digital access, then they’ll need to speciate into something else to survive the emergence of CBDC’s.

DAI perhaps has a slightly different angle. It’s collateralization algorithm requires a deposit of more than 1 Dollar to generate 1 Dollar, but it allows this from all manner of assets, and maintains a stable price throughout the trading day. This means that, if you’re worried about minute by minute fluctuations in ETH, you can put it into DAI and use that as a genuinely decentralised payment stable. It’s replete with all the other things you’d expect to see in what’s basically a current account - a savings rate, the ability to unlock excess capital if fortuitous transactions arise in whatever you bought the DAI with, so on. The collateralization algorithm will put a lot of people off, especially those who’ve never been stung by high gas fees or a low daily price before, but as people get more experienced with the space, DAI may have a genuinely well speciated use case which helps dollar natives translate into Web3 and survives past CBDC introduction for those who like their finance decentralised. DAI just needs a translator. And, for the others to not solve their volatility problems…

When Freed Under Stress, A Tether Will Snap

To repeat myself ad nauseam, 64% is a big market share to suddenly come under increased scrutiny. If that scrutiny goes the way most of those in the know think it will, then some kind of replacement will need to emerge, most likely from extant competitors. I’ve made the potential use cases out here; what happens next is largely dice roll and a matter of what’s actually behind Tether’s apparently incomprehensible curtain (which, by the way, is never a good defence; Ignorantia juris non excusat).

I’m more concerned that we’ll see increased demand for regulatory oversight and, more disturbingly, manual control, in spite of CBDC’s doing 10 times this, if Tether’s downfall is handled poorly by the space. And, if CBDC’s are coming in as early as 2025, we can’t afford to lose stablecoins to state control. I continue to maintain that the whole SBF saga was a strike against regulation, not in favour. A malinvestment was made, a malinvestment was corrected. There was some retail loss.

Get over it.

That’s what markets do. And their freedom is more important than marginal retail loss. But, for the Web3 vanguard to preserve that for those who come after us, and for our customers, we need to act responsibly and get the lifeboats out.

Now.

Stratified Costs

Many RWA projects face internally stratified costs. In the case of commodity or precious metals; it can be more difficult to bring one component part of the same combined asset to fruition; ground gets hotter and ores get denser the deeper one digs. One can straightforwardly imagine a way to arbitrage excess demand into supply of such a component, the same way any crowdfunder goes to market and says:

‘Hey, please give me money for x, you’ll get y in response [ahead of time]’ even if y is just feeling charitable.

But scaling this model, especially under CeFi, has limited flexibility. Transaction fees are sickening, high value transactions have to be individually approved by law, conversion into more advantageous currencies wherever you are is expensive; everything works worse, but is entrenched and so is defaulted to.

Moreover, CeFi, by its declaration and disclosure requirements, promotes the sorts of perverse incentives which set up oligopolies in exactly the sort of new markets we want to be competitive. How?

Well, remember those higher costs? Those are already raising barriers to entry, and, more importantly, promoting strong selection biases for those who can get past them to economise at scale. Combine this with the knowledge that you can only get as far as one disclosed jurisdiction plus its connections before competitors jump on you, and it doesn’t make sense to run the completely honest go to market which these economic models are fundamentally predicated on, especially if I’ve reached the point where I don’t need to do this. Too big to fail, too grandiose for candour. It instead makes sense to channel your inner Julius II, and Treaty of Tordesillas yourself and any competitors into strictly defined zones where there’s less competition and the consumer ultimately loses.

DeFi can fix this both for the ambitious producer and, by extension, consumers, even if they never notice it. I’d almost go as far as to say the consumer not needing to focus on their financial system is a compliment.

Tokens As Rights

Imagine that some variation of the Green New Deal, mandatory ESG, so on so forth, actually becomes law. I own a Lithium mine, and demand goes stratospheric because now all those additional batteries which have to go in those electric cars need my commodity to work profitably. However, as above, parts of my deposit previously didn’t make sense to exploit, and, even with this newfound sovereign backed demand, some smaller part still doesn’t. We now have three tiers of stratified product going to the same ends. CeFi doesn’t allow for easy arbitration on this, never mind elegant expression of how that cost is going to vary.

Now, imagine that I issue a token instead, which, upon certain requirements being met, entitles the holder to the produce, or profit equivalent of, this mine, corresponding to the volume of tokens they hold. The advantages for me are obvious; I’m not locked into long term instruments, I can market for liquidity much more widely, and I’m not subject to CeFi mission creep. I can also market all of these things as points to investors and consumers alike.

On top of that, I can treat the different strata of my product with bespoke tokens, or change the release order to simulate that difference, or do any one of a hundred other things. Importantly, I have an immediately flexible way to respond to demand, because demand can now be more directly signalled. If the globe gets hotter than expected and regulations tighten further, I can open up operations into that third strata with code in seconds, whereas a parallel operation in CeFi would involve vis maior clauses in the innumerable contracts I’ve signed, which go to any number of arbitrators, so on and so forth. Much more costly for everyone involved; far more distorting on the actual price consumers end up paying, especially where demand inelastic goods are concerned.

Less obvious, and far more powerful, are the additional benefits to investors. Like me, they aren’t locked into long term financial contracts. They can trust the code, rather than eleven different bankers and all their incentives. There’s now, just as in any other DeFi space, a market on those tokens. You don’t have to care about Lithium to get involved in my venture now - you can directly speculate strictly on the financial returns without any CeFi restrictions getting in the way.

This then allows all manner of interested and specialist parties to get involved in their own way; yield farming, options, ETF’s; we can directly plug high fixed costs projects into the liquidity markets they need to get prices down and wealth in real terms up. Human deception and error can be eliminated because, code is the law here. Tying back to the above, the lack of disclosure requirements reduces what my competitors know about my capabilities and ambitions. It’s now decision theory, rather than general game theory. They’re forced back into behaving like exactly that, competitors; I’m forced into actually carving out what I want based on product quality and customer service, rather than cornering the supply ahead of time and maximising for my own surplus.

I can feel Hayek smiling from here.

Come One, Come All

All of the above is true in the case of one commodity with a specific range of use cases and a variable cost. However, mines aren’t pure. Going back to my example, Magnesium is often found alongside Lithium. Like Lithium, Magnesium has a range of commercial use cases. If its lightweight and electric, or needs to maintain internal stability at high speed, it probably has Magnesium in it, but notice that these use cases are subtly different from Lithium. Having enjoyed my runaway success tokenising Lithium, and noticing the low XED in my products, I want to tokenise Magnesium too.

And now comes the cherry on top; liquidity pools. DeFi, can, by setting up a pool of tokens representing my Lithium supply compared against tokens representing my Magnesium supply, do more than eliminate extortionate CeFi currency arbitrage.

It can directly democratise demand for any comparably pooled asset.

This rests on some assumptions, but not necessarily the ones you’d intuit. My ownership of the mine at this stage doesn’t matter very much - once the tokens are out in the world for exchange, as long as the mine keeps mining, who owns it matters very little. The core assumption is that whatever rate of supply all this overlay infrastructure was posed on remains inside tolerance limits. This is, as with every other financial instrument, more true as you rack up risk. The heavier I bet on Lithium against Magnesium, or vice versa, if supply of one stops, I’m in trouble. But there’s no reason for this risk to become existential to the given DeFi system; much less so than in CeFi because, again, code is law. There’s no incentive for larger agents to conspire to push the leverage beyond what’s sane, knowing that the taxpayer will backstop them - there’s no taxpayer here.

This also allows consortiums of smaller retail investors access to profits reserved for institutional actors under CeFi. Say a well-performing retail investor gets news of incoming regulation; my Magnesium extraction operation goes so well that I’ve flooded some electronics manufacturing market somewhere, and the price is going to fall imminently, causing some other tokenised and pooled asset to rise in response (which need not be Lithium at this point). Well, they can take out a flash loan and arbitrage on the price hike on the other commodity - easy profit, directly to consumers.

C.F. Fiat Stables

Comparing against fiat stables provides further elucidation. Counterparty risk is (largely) ameliorated by using a commmodity - as long as there is liquid demand for the underlying commodity and confidence in the token, there’ll be someone looking to purchase. This is not the case where underlying fiat positions are being abandoned or rolled back. Some others would have been in the worse position of factoring their structural inequalities into their distribution, but MiCA more or less kills any possibility of a competitive European stable.

A niche, but perhaps, dare I say, dull market in the underlying commodity is advisable; those in broader demand or with more dangerous applications are likely to have their movements around the globe regulated and tracked - all complications project directly onto any pegged token. This must in turn be reconciled with logistical constraints and liquidity concerns - the backer can be as confident as they like, but the stable is still limited by the total market of the underlying commodity. Perhaps some fractionation is in order - it’s difficult for me to ship rare earth metals around the globe on a whim, but if I incorporate them into a larger order of more common deliverables, average prices start dropping. Similarly, if my token (and, chemistry allowing, physical currency to match) is based on a rarer commodity to the extent that a price is readily derivable, I can, up to my established risk appetite, build that into something more commonplace to ease liquidity concerns, and effectively optionise liquidity.

I mention physical coinage representing a token as another approach to liquidity problems, especially ones related to friction and uptake. If it becomes just the next kind of cash, the average person can use what is also a stable and a token from their physical wallet, as well as a chain one. All that is necessary is drumming up the underlying capital to match the token with parallel coinage, and convincing whichever central banks one intends to target that introducing this parallel coinage won’t have disastrous effects. An awful lot certainly, but, a visible amount. This is perhaps the most heterodox ambition in this paper, but in a deglobalising world, I remain hopeful. Those who choose to break from the dollar over pride, prerogative or prudence could find a cooler political alternative in a decentralised, private option, set apart from centralised geopolitical intrigue and diorama. Especially in jurisdictions where cash substitution is a long running issue - none of the political friction of CBDC’s (nor expenses incurred in building the necessary infrastructure on the public purse), easy integration into all extant markets, and direct tracking, assuming some mechanism to correspond a hybrid coin to its token (the most difficult part to build into a technology which would try to solve both these problems at once, but necessary lest double spending take hold everywhere). An ATM like dispensary connected to a wallet might be a good start.

But, drawing our metaphor to a close: if I don’t like how separate my Lithium and Magnesium markets are, even if I can address them separately, and tokenise those efforts, how about I fuse the two with a token made of 50% Lithium, 50% Magnesium? Ok, don’t put it in water, or heated air, or pure oxygen… but you get the point. Liquidity (or reactivity) concerns? Just dilute into a different ratio with a parallel refining product and resume the process.

Especially if favoured counterparties could be made ready, waiting to provide liquidity across the barrier until it dissolves, and then translate customers into the new paradigm once I decide to make rid of it. Instant market-wide efficiency improvements, and wider uptake appeal.

I’d have triangulated demand between, and thereby profit maximised the supply of, my individual products, derivatives thereof, and fusions therebetween, in one fell swoop. Oh, yeah, and created a parallel financial ecosystem, limited only by scale and adoption.

Aut Caesar, Aut Nihil.

Concluding Necessities

It’s not obvious where such a system should be based or begin from. Stables would make a good place to start, and neither the EU nor US is a good home for them right now; this will be further commented on later.

Further, as you’ve likely surmised, it’s likely to make CeFi, and all its institutional power, very, very angry, long before DeFi figures out its advertising and comprehensibility problems. In an uncertain, de-globalising world, only relatively insulated and uninvested jurisdictions make obvious homes, and the US isn’t a good place to start as long as the SEC maintains its current course. Mexico, Canada and Australia might all be strong candidates with commodities connections and some kind of shock insulation from changes in trading markets.

However, for the ambitious, I can think of no better triple offensive to legitimate DeFi, establish a legacy which spans the transition from CeFi into DeFi, and which garners prestige to last generations. All remaining notes concern judicious deployment; far better handled by operatives than ideates.

Scaling Stablecoins

Stablecoins are easy enough to understand in and of themselves. Just as one currency may be pegged to another, stablecoins are cryptographic tokens which have a value pegged to an underlying asset or series thereof. Some are simply pegged to traditional currencies like the dollar, allowing a 1-1 exchange ratio for the consumer. Note that institutions may not treat them as 1-1 equivalent due to differences in confidence, storage, regulations, risk appetite, and so on, but deviations will be small. Additional complexity can be added - some stablecoins could be pegged to stock indices, commodity values or a basket combination thereof. The Stablecoin thereafter has value both as a medium of exchange and a store of the confidence in the value it is pegged against.

This allows those who mint stablecoins, presuming that they also hold the underlying asset, an unparalleled opportunity to develop their own influence and propagate their designs into the future. The first challenge will be ensuring that the stablecoin is fungible in terms of some other currency, or that it has purchasing power in real things. This can be in terms of other currencies like the dollar, or goods and services, in the way Bitcoin and Ethereum increasingly are increasingly accepted as payment directly to merchants. Ideally, the minters, or consortium thereof, would aim for both simultaneously, and pursue whichever avenue posts the most returns more aggressively in-line with their broader strategy.

Financial Potential

This does not mean that the underlying asset is now off the table for further deployment. Certainly, one could use a certain amount of underlying commodities to produce cash equivalents, physical representations of the token in question. That commodity could be combined with other, more abundant supplements if it is particularly rare and/or difficult to physically manipulate - a composite coin of say 95% scrap metal, 5% value metal can still function as a Real World Token (RWT) for a stablecoin based on the value metal, just the same as one purely composed of the value metal, because it is a token, rather than a lump commodity.

If a value metal were to have other applications, one could therefore collect money on it twice. Creating a stablecoin pegged to its value is the best of both worlds in that it is a store of value like gold and Bitcoin, but also uses more sophisticated methods, and has immediate real world appeal. What remains of the value metal deposit not deployed in real world tokens could then be used in any other applications, and sold to those who would so use it. It goes without saying that, if the minters of this stablecoin, were to control a large fraction of the global supply of the value metal, integrating themselves into the supply networks using that value metal would afford them volumes of soft power and prestige, alongside that already gained from creating such a desirable currency.

However, if the minters were to try and pursue both these options, exactly what to peg the stablecoin to becomes a more complex decision. If a significant fraction of their total supply of value metal is being sold elsewhere, pegging to the global supply of the value metal in question becomes less stable, as the buyers cannot be controlled once they purchase, and may fail in their own enterprise, deviate from contract, or produce other unexpected results. All these things would presumably influence the supply value, unless sales of the underlying value metal are so wide that any one actor no longer matters on the demand side. Whilst this might be the case in retail and wide use metals, I cannot envision this easily being the case with the Rare Earth Metals which will likely be used for stables in this way. It may instead make sense to peg the value of the stablecoin to an entity which holds:

• The initial ecosystem and blockchain on which the stablecoin is to be minted and deployed.
• The deposits of the valued metal in question.
• Contractual interest from those looking to acquire that valued metal.

This is a less subtle approach which would require some degree of interaction with conventional financial architecture, though likely on the supply side. This may in turn compromise anonymity to some degree, if this is important. However, note the specific word, ‘hold’. There is no requirement for the entity in question to actually own the assets in question, or derive any final beneficial expectation at law. The only requirements, should the minters wish to pursue this route, are that:

• The entity is an effective confluence node for all these derivations of value to intersect.
• Those holding positions in the entity have some way to derive benefit to their satisfaction from the entity (resulting trusts, debt and licensed intellectual property are all routes by which this could be straightforwardly architected).
• That this architecture is not so effective that hostile intentions arise from those in Centralised Finance (CeFi) and elsewhere whose business model and interests are disrupted, at least, outside tolerance or ability to ameliorate those intentions.

Enlightening Economics

One underrated point of Hayek’s economic vision was a system of competing private currencies, sometimes complementing, and sometimes set against, Von Mises’ bet on gold, now going digital. Cryptocurrencies as currently formulated go part of the way towards fulfilling both these visions. Schumpeter further noted that radical breakthroughs will be followed by smaller clusters; the first mover on this technology would be perfectly poised to make such a breakthrough which would go down in history.

The first breakthrough such a project could achieve is the reconciliation of Wieser and Von Mises. Wieser predicted that currencies would degrade, and separate from their subject matters - Von Mises subsequently insisted that money managers would not strike in the interests of the citizenry and that ‘gold must be the currency of everyone’, so that they always had access to real value which they could lever and depend on. Bitcoin is a store of value analogous to gold. But it will never achieve currency adoption and status on its current architecture. A Stablecoin could achieve this, whilst maintaining the ability to store value like gold and Bitcoin.

The second is advancing the view of money as a comparative tool put across by Schumpeter, far ahead of his time. Schumpeter uniquely appreciated that money is still important where no transaction takes place, because it allows for an easy comparison of the value of goods, and for individual agents to calibrate their future behaviour accordingly. There must be a common unit of account for trade to take place, because meaningful comparison implies a standard to be measured against. All this is true at one point in time; when iterating forward into the future, it is best for this standard to be as stable as possible. This is where stables come in. Government monetary and fiscal policy, whether they like it or not, manipulates the value of the standards of comparison their citizens use nonstop. People’s individual tendencies to spend or save are exacerbated by inflationary events and decisions beyond their control. National fiat currencies may be widely used, and preferable to barter, but they are also sub-optimal standards for measurement. Stablecoins, proof from direct government engineering and separate from political incentives to debase the currency for the next big spending project could provide a far better standard to measure goods against as a general ledger.

The final question remains one inside the space - why use stables over a currency confined to a platform, and why are they preferable inside the Web3 space to other cryptos? Whilst larger firms may have the power to profitably issue their own currency and accept only it as legal tender to better set up price discrimination in their favour, smaller firms will have to use common units of account. Even then, it doesn’t automatically follow that larger firms will choose to issue their own currencies, as it may dissuade customers who can easily go to a competitor. Either way, once we’ve accepted that there must be a common unit of transaction, stablecoins should be easily engineerable for low gas fees, based on the ability to specialise them for exchange, excluding development tools for other purposes.

This way, they could be the reconciliation between Schumpeter and Hayek; Hayek famously argued for private currencies to compete in the same way cryptocurrencies do now, but Schumpeter worried that winner takes all dynamics would lead to a net loss for consumers on account of exactly the price discrimination previously mentioned. With stablecoins, the remainder of the crypto ecosystem is free to compete to build the most efficient currency possible, but, with a separately indexed price and value, consumers do not stand to lose from being individually price gouged based on their position, buying habits and demands.

Conclusion

We feel that the intersection between RWA and Stablecoins is one of the most underrated points in the modern crypto ecosystem. Sentiment is currently elsewhere, chiefly focused on CBDC’s, with most commentariat attention migrating to AI for the last few months. There is, in this distraction, a chance for a consortium to be first to market on an entirely new class of currency and take not just market share, but huge volumes of soft power by storm.

This would be best done by creating a stablecoin ultimately pegged against, if not a valuable underlying asset, an entity in a position to corner the supply of that asset and monetise its supply. Two for the price of one. This will achieve a tripartite offensive: first into Web3 by creating a universal currency and store of value, which also has an intuitive appeal current cryptos lack. Next, into whatever markets the underlying asset fuels, building the entity and its holders huge soft power, alongside a real-world ecosystem to deploy the assets their stable is backed against. Finally, the intersection between that supply and Web3 will, correctly managed, produce a fusion of underlying assets and alternative enterprise, woven under a financial umbrella, which could ground the expansion of that stablecoin into ground normally occupied by CeFi. When it does so, it will not be as capital vulnerable as Bitcoin or Ethereum, in financial or social terms, because of the huge volume of underlying assets and depending processes.

An unprecedented opportunity awaits. Qui audet adipiscitur.

Foreword

For the majority of 2022, I worked at a Web3 wallet startup that aimed to transform how we interact with Web3 in its current form. There were clear challenges that we specifically went after, and hence born the moniker “user experience infrastructure startup”, given everything we did was catered around solving very clear UX issues.

Fundamentally however, we were a wallet startup.

A wallet that solved the key management problem. A wallet that solved the gas-cost interaction problem. A wallet that solved the ramp problem. Simply however, just a wallet.

There are lots of people looking at one or some of these challenge and from different angles, but for a really complete experience, we knew that whatever the final product was it needed to be seamless; central to the customer experience, and everywhere.

It is from this base position, and conversations I’ve had around it, that I’ve refined my thinking. If you haven’t already, I thoroughly recommend the “Wallet Centric Customer Experience Stack” by Ben Basche. Recent conversations with him have only reinforced my views here.

Identities

What is a wallet? At a base technical level it can be as simple as a public/private key pair, or as complex as an on-chain smart contract. At a philosophical level however?

It’s a key to your digital identity. An identity you need to do anything online.

Digital identity is something that we’ve been exploring for decades now, even long before the first smart contracts were written. It’s come in many shapes and forms; from the humble username and password, to the behemoth that is OAuth.

Cyber security professionals have wrestled with identity since cyber was a thing, modelling the best way to place who someone is, and what they can do, with some models being more successful than others. These models were all incomplete.

It turns out that your digital identity and your financial identity are in fact inextricably linked. We have built billions of dollars of products and infrastructure around this key fact. Paypal, Stripe, Apple Pay; all products that attempt to wrangle these 2 separate identities into something a little less painful to manage.

Consider the customer experience of almost all sites these days. You’ll likely login with Google or Facebook, proceed to find the thing that you want, be it a service or a product, and then go to some kind of checkout flow. This checkout flow will allow you to select from one of the integrated payments solutions, be it something more seamless such as Apple Pay — which has gone to great lengths to attempt to merge these 2 identities — or more traditionally built solutions such as Stripe checkout. Removing from the equation the heavily convoluted settlement layer that is our traditional financial infrastructure, it’s pretty clear that really the most desirable experience here is one where all of these payment methods come with your digital identity;

I’d posit that if “Login with your bank” was an option because they pioneered OAuth alongside Google, it would have been incredibly popular. You’d already be KYC-ed if that was necessary, payments would “just work” and would be secured using the same teams that keep your money safe in the bank.

Experiences

Outside of revolutionary changes in Human Interface Devices such as the newly released Apple Vision Pro, the base experience apps should have is a pretty solved problem.

Pull down to refresh. Swipe to go back. Single click login gets you to the start of the value chain.

It’s been made abundantly clear that the seed-phrase based approach to wallets thus far is completely unfit for purpose in that regard. Users don’t want to have to manage their seed phrases, no more than they want to manage their own email server.

The Wallet Centric experience starts here, with a creation and “login“ (read: connect) flow that looks and feels almost identical to the login flows we have today, getting users to the key value of your product quickly and with low friction.

This isn’t a particularly out-there realisation. Several of the biggest names in the space — Uniswap notably among them — have decided that the “embedded wallet” native within their mobile app experience is the way forwards. These approaches work for getting a larger wedge of market-share by capturing those who churn due to entry costs; good for their short-term usage numbers, but eliminates key parts of the value of a true wallet-centric experience.

I don’t have multiple identities across multiple apps where each app has created a key-pair for me.

I have one identity (perhaps with N personas, more on that later) which I want to project into all of my digital experiences.

The contents of my wallet should allow my experience to be tailored. This can’t be done when its fragmented across multiple wallets where I would have to export the private key to move it around.

Ramps and Rails

Halfway through writing this piece someone sent me Simon Taylors Future of Payments blog; in there being a particularly pertinent comment;

India, the home of UPI and digital identity, doesn’t yet have an open banking standard, which suggests identity is the proper foundation of payments if you were to start again. No?

Crypto can act as payment rails, a point Simon also calls out, but independently it can act as a custodian. If I want to access DeFi pools with above average returns or escape my countries currency volatility and hedge it by simply holding stables, I want to have a combined digital and financial identity that can allow me to do this.

My previous Letter to DeFi talks about how the future of DeFi will look a lot more like a modern Neobank than it does what we have today. There I also touch briefly about how if you were to build one around a Wallet Centric experience, you could create something seriously compelling, and indeed there are some players barking at this thesis.

Behind this experience you can then begin to abstract away the ever increasing number of payment rails that are being created — and the crypto ramps that want to service them — reaping the benefits that these teams are creating in attempting to create smooth fiat rails for customers as you float between the 2 systems.

A value proposition that can now be brought to all apps; a single contact point for developers to build from and a single identity for users, fiat and crypto built in.

As a passing comment I suspect this is Coinbase’s to lose. They have an exceptionally strong position in being one of the biggest on-ramps into crypto at the moment — one that’s also attempting to do it properly no less — and are actively pursuing a “Wallet as a Service” strategy. I highly suspect that the natural evolution of centralised exchanges will converge to banking which leaves Coinbase; then a bank with a digital wallet identity and deep token liquidity, sitting in an incredibly strong position. I digress, an essay for another time.

The Loop

The future of crypto/”web3”/whatever will have users enter because they arrived at an app they wanted to use, ”logged in” and proceeded to have an identity created for them.

Where this has been done through a wallet-centric approach, it’ll be this same identity they have when they arrive at the next app they want to use. The cold-start problem here is ensuring that apps want to build from a particular wallet because they solve genuine user challenges such as rails, gas, or chain complexity.

The loop arrives when apps want to build out from this wallet because they understand that this where the users already are.

Merchants accept specific payment methods because they want to attract and capture the users that are found within those systems. The future of digital identities will have apps building around the identity instead, knowing that with it comes the financial identity with the relevant rails because they’re now one and the same.

It’s here that once the base experiences have been solved, that the “identity providers” will begin to differentiate.

Providers can open up more APIs to developers to tailor their experiences further, be it from new loyalty offers, to financing options or even cross-chain. Alternatively the experience might be more straight-jacketed, ensuring that the experience across apps is always reproducible and understood, attracting a host of users who churn at too much choice and want to simply access a very simple and refined digital experience. Winners here will, as ever, be the ones that can identify what users and builders want and iterate to get the right product market fit the quickest.

Whichever provider masters the “personas” challenge, allowing me to have uniquely tailored experiences depending on a certain identity I currently want to use, but without having to “login” to entirely new systems and break out of flow, stands a strong chance of creating a very defensible base offering.

Closing

Our current digital identities are badly overdue an overhaul, but simple changes over the “Login” flow are insufficient to realise the true potential here.

How digital identities should feel is a solved consumer problem. No-one but the maxi’s insisted on having to manage your own private keys, and their time is short lived.

How financial identities should behave is a solved consumer problem, with minor iterations enabling new experiences in certain spaces but build on fundamentally the wrong primitives.

The convergence of these 2 identities will enable all users to project themselves digitally in the way that they’re supposed to, not missing out on certain experiences just because they made the wrong consumer choice of what card issuer they went with.

The Backdrop

Decentralised finance, shortened to simply DeFi, is the fundamental notion that access to financial services and security should be a universal right, and that we can leverage decentralised technologies to be able to realise this idea.

Blockchain and crypto technology can be used to bring us products that allow us to swap currencies and other tokens, safely store digital value, trade and exchange this value, and earn a return on that value through various mechanisms, all outside of the major centralised institutions and without needing any explicit permission to use them.

Current State

The current state of DeFi is frankly, an absolute mess. Plain and simple. The blame here isn’t all for DeFi, the crypto industry in general is in need of maturing right across the board.

Currently most of the volume and product launches are still being targeted towards speculators and a liquorice assortment of other bad actors. I’m not saying speculation is intrinsically bad, but if this is all this industry amounts to be, then that will be exceptionally disappointing.

Crypto in general is suffering heavily — near mortally — from a user experience problem. The simple idea of getting a wallet, buying some tokens, trading them, paying for something, or other actual useful applications of the space, is obtusely challenging, even for tech literate people.

Genuine, sector-changing attempts to tackle these problems are being hemmed in by ideologues and other maxi’s who think that “oh but if its not completely perfectly decentralised or trustless or blah then it’s clearly bad”. It’s time for them to grow up.

I’ll be writing a whole other piece around the “principles vs mass adoption” argument so I won’t get into it too much here, but at a base level, it’s insane that I even need to argue it. The time of the gatekeepers is coming to an end.

The Need

The technical capability to drive serious change is already here. I was part of an elite team last year that spent their entire time dedicated to solving these problems; and we did. There are plenty of teams out there that are seriously trying to drive the needle forwards, but hemmed in by this idea of “this isn’t where the market is right now” and in some sense thats true, the mistake being that if they think just servicing the current load of speculators is the only market there is, they’re wrong.

If we break down the world of consumer finance, it becomes pretty clear what the market has decided it wants:

• Store my money safely and let me spend it where I want to, how I want to, and don’t charge me too much to do that
• Let me save some of this money and maybe generate a return on that. If you could outperform inflation, that would be awesome.
• Let me send money to other people easily and safely (not safely comes second, as CashApp demonstrated)

If you ever find yourselves in the upper echelons of the financial consumer, you might also want:

• The ability to take loans and other credit products
• The ability to unlock liquidity in certain assets, such as wanting to refinance a house

Currently it is incredibly non-trivial to be able to do these things in DeFi, though thats not to say the primitives don’t exist. There is infrastructure and products out there that if you squint at looks almost exactly like this, but using it is basically impossible;

• There are lots products around spending crypto online, and in-person. These are particularly rampant in Latin America where the desire to get away from their native currency systems is palpable.
• You don’t have to go very far to find mechanisms for earning yield in DeFi, even when you strip away the ponzi schemes.
• Sending tokens between two wallets was literally one of the first use cases for crypto, just doing it isn’t a very enjoyable experience.

And with recent innovations we’re able to find;

• Platforms for unlocking liquidity in assets such as NFTs
• Real World Asset (RWA) platforms that are attempting to bring things like real-estate on-chain though these are very much still in their infancy

That Latin America point is quite poignant. Those of us that live in The West have this warped view of the world about what crypto really is and and how its being used, but throughout emerging markets, you’ll find real people who use it day-to-day to exchange money for regular household goods because even this hostile experience is worth it to them. They deserve to be serviced better.

The Vision

The future of DeFi will, to consumers, look a lot like modern Neobanks than they do their current form. There will be more than 1 player. They will choose to differentiate in different ways. This is the sign of a mature market.

They will be “wallet centric” apps similar to Neobank apps but not quite like how they look in either of them today. This app will be a gateway and key into a host of new apps, but you’d never know it; it’ll be cleanly abstracted away. This might be because the wallet provider itself, and the bank provider are separate entities, it might not be. The market will work out which of these models will work (more on wallet centrism in another post)

The app itself will be denominated in actual native currencies. Wallets denominating on ETH is ridiculous. ETH is not a currency, SOL isn’t a currency. USDC is a currency.

Users will be able to spend their money using traditional rails, and new ones. There will be debit cards that work on traditional rails, and products that manage this complexity will have strong moats.

We can transfer money to raw addresses, resolvable addresses such as ENS, or other KYC-ed pieces of data where they want to such as phone numbers or emails. This KYC data might be on-chain but private, it might not be. Either way the consumer market will require trivial mechanisms for the transfer of their money between their friends, and the market will provide it.

You’ll be able to recover their accounts. Trust will be put in entities or mechanisms that are considered intrinsically safe. For those who choose not to have this luxury and insist it must be trustless, there will be mechanisms for them too, though the complexity will increase accordingly. This will be okay, but won’t be the norm. The gatekeepers and maxi’s will shout, they are to be ignored *cough, ledger, cough*

A Cambrian explosion of new types of “savings accounts” will arrive, powered by the permissionless on-chain primitives of DeFi, finally unlocking its true power. Billions in liquidity will pour into pools, tokenised asset classes, project funds to bankroll initiatives people actually care about with verified returns, and all without needing to hunt them down in a fractured online nebula, an exercise we know most users simply do not care about.

People will be able to see completely how their money is being a force for good in the world, the carbon its removing, the charities its helping, the small businesses that are growing. It’ll be tracked and rewarded on-chain, and with it a new host of consumer benefits will arrive, with brands looking to target their specific niche of customer, creating new go-to-market strategies, and consumer choice.

Money shall be transacted. Preference shall be signalled not presumed. Experiences shall be tailored not modelled. The birds shall sing, and the sun shall shine.

Landscape

There is currently a handful of players who are all barking at this thesis, but there is still such a long way to go. A few honourable mentions I’ve been tracking;

• Ultimate. These started out as a DeFi specific wallet, which naturally falls under that whole “servicing the current speculators” thesis, but are chasing down the Neobank play. It’s an interesting GTM if they can execute it right having already captured the speculators, now can they service the regular consumer.
• Meld. These guys went the other way, building out a clear Neobank-on-top-of-wallet play first, and have done a pretty polished job of it. If they can execute an effective GTM and product roadmap, could be curious to watch
• Privy. An execution against “Web3 wallet for all the users”. They’re shaping up nicely to capture the mass-adoption wallet market. Should this be coupled with the right DeFi experience; could easily be a contender to bring the DeFi revolution we deserve.

I’m sure there are others, and I’m sure twitter will have fun pointing them out to me.

The Institutions

Institutions are not taking this lying down. Almost all of them are exploring what this means to their businesses; I talk to a lot of them. They, however, have more explicit and clear requirements of what they want and need out of the space; and what they’d be prepared to tolerate to get into it, something we’ll save for another post.

As the biggest names in traditional finance have declared that “tokenisation is the future” so too have the lines in the sand been drawn; this isn’t going away.

As TradFi bring their massive portfolios, risk appetites, and debt servicing, partnering it with the efficient, liquid, and open markets that DeFi provides, we’re destined to enter a new golden age of financial products for retail consumers not just in the west, but across the world.

Thinking that the sector itself is somehow better off without their interest or involvement is just silly.

Yes DeFi was in part born out of the notion that traditional finance has failed us — and in many cases it undoubtably has — but for it to really mature into the revolution it deserves to be, it’s time to put down the pitchforks and torches and have sensible conversations about what that looks like.

The regulatory scene is going to change around crypto, in the same way that it will all of finance. This pipe-dream that you can just run around in the crypto industry with no care or interest for consumer protection, international law, and basic financial prudence, is exactly that, a pipe dream. If this is you, grow up.

This isn’t to say that DeFi won’t be revolutionary change, it will. Being borderless and permissionless will allow global financial paradigm shift that means everyone everywhere can access the financial primitives they deserve, in a manner that doesn’t require you to have a masters degree in computer science to be able to use it not get scammed along the way.

Closing Remark

It’s time for this sector to evolve on-the-whole. It’s time for the gatekeepers to be moved aside. Not just so that we can realise the full potential and vision of DeFi, but so that the innovations we make along the way can unlock whole new ways of living and structuring society. The market has clearly signalled how it is prepared to interact with the digital world, it’s time we come and meet them there and service it.