For centuries, liquidity has been a human endeavour and to a large extent is the reason why money exists - purely bartering for goods and services is more complicated. Traders, institutions and market makers have all played their roles in ensuring that assets can be bought and sold efficiently; even in the early days of digital assets, liquidity depended heavily on human decisions, where to allocate funds, when to provide depth and how to respond to volatility. In today’s digital asset markets, capital is beginning to act on its own. Not metaphorically, but literally. Algorithms no longer simply execute human-defined strategies - they are increasingly defining the strategies themselves. The emergence of artificial intelligence in financial systems is giving rise to a new kind of participant: one that does not sleep, does not hesitate and does not require human input to operate.
Flash Crash
Source: Amazon
To figure out where we are going, it helps to understand where we have been. Traditional liquidity provision was dominated by centralised actors, banks, asset managers, market makers and proprietary trading firms. These entities employed teams of traders and analysts who manually adjusted positions, set spreads and responded to market movements. Liquidity, in this context, was both a service and a strategy, shaped by human judgment and constrained by human limitations. It was not until the late 1980s/early 1990s that algorithmic trading (featuring fully electronic trade execution) made its debut in financial markets. A pivotal moment came in 1998 when the US Securities and Exchange Commission (SEC) approved electronic exchanges. This regulatory green light paved the way for the rapid rise of computerised high-frequency trading (HFT). Being able to execute trades up to 1,000 times faster than any human, HFT quickly became widespread and transformed market dynamics. Yet, even then, the intelligence behind these systems remained largely static. Humans designed the models, machines simply executed them. The next evolution arrived with decentralised finance (DeFi), where automated market makers (AMMs) replaced traditional order books with mathematical formulas. Liquidity was no longer provided by institutions alone; it became a programmable resource to which anyone could contribute. This democratised access, but it also introduced inefficiencies. Static liquidity pools often struggled to adapt to rapidly changing market conditions and now we are entering a new phase, one defined by adaptive intelligence. AI-driven systems can analyse vast amounts of data in real time, identify patterns and adjust strategies dynamically. However, unlike traditional algorithms, they are not bound by predefined rules; they learn, evolve and optimise continuously. In this context, liquidity is no longer simply automated, it is becoming intelligent.
Autonomous liquidity pools
Imagine a liquidity pool that does not require human intervention to function. It is created by an AI agent and is funded by capital that has either been allocated or has autonomously been generated through prior strategies. It monitors multiple markets simultaneously, reallocating assets based on volatility, trading volume and cross-chain opportunities. If volatility spikes in one market, the pool may withdraw liquidity to reduce exposure. If another protocol offers higher yields, it may redeploy capital accordingly. If an emerging asset begins to attract attention, it may allocate liquidity pre-emptively, positioning itself ahead of the curve. And when the opportunity disappears, it shuts itself down. This is not science fiction; it is a logical extension of existing technologies. Smart contracts already enable programmable capital. Cross-chain infrastructure allows assets to move freely between ecosystems and AI provides the decision-making layer that ties everything together. The result is a system where liquidity is no longer static or reactive but dynamic and anticipatory.
Moreover, these autonomous pools could operate as independent economic agents, competing with one another for efficiency and profitability. Over time, the most effective strategies would accumulate more capital, creating a feedback loop where success breeds scale. In such a world, the role of human liquidity providers diminishes. Instead of directly managing capital, humans may simply deploy it into AI systems and allow them to operate independently. Unlike traditional financial markets, they operate 24/7 with no closing hours or downtime and hence, one of the defining features of digital asset markets is their relentless continuity. However, humans don’t function this way and, unsurprisingly, this fundamental mismatch has long created inefficiencies. Market opportunities emerge at any hour of the day or night, yet human traders are limited by attention spans, sleep cycles and personal availability. Even conventional algorithmic systems, whilst extremely fast, typically rely on rigid, predefined rules that struggle to adapt to complex, rapidly evolving market conditions.
AI changes this dynamic entirely
An AI-managed capital system does not merely execute trades, it interprets the market in real time. It can monitor funding rates across derivatives platforms, identify arbitrage opportunities between exchanges and assess liquidity imbalances across chains simultaneously:
- When a pricing discrepancy appears, it acts instantly.
- When sentiment shifts, it recalibrates.
- When volatility increases, it adjusts exposure.
The speed at which these systems operate fundamentally alters market structure; price inefficiencies that once persisted for minutes or seconds may disappear in milliseconds. Liquidity becomes more responsive, but also more competitive. In this environment, the concept of “edge” evolves. It is no longer about possessing better information; it is about having better intelligence. However, there is a paradox here. As more participants adopt AI-driven strategies, the advantages they provide may begin to diminish. Markets could become hyper-efficient with fewer exploitable opportunities. At the same time, the interactions between multiple AI systems could create new forms of complexity, emergent behaviours that are difficult to predict or control. One of the defining features of digital asset markets is their relentless continuity. The idea of autonomous capital is compelling, but it is not without significant risks - one of the most immediate concerns is the potential for runaway algorithms. When AI systems operate independently and make decisions at high speed, small errors can quickly escalate. A misinterpreted signal or flawed model could trigger a cascade of trades, amplifying volatility rather than stabilising it. Traditional markets have already experienced flash crashes caused by algorithmic trading - the 2010 “Flash Crash” offered a dramatic warning of how algorithms can interact in unpredictable and destructive ways. On the morning of 6th May 2010, more than $1 trillion was wiped from the value of US equities in just minutes as automated trading systems triggered a self-reinforcing spiral of sell orders. Yet, despite years of official investigations, the exact cause of the crash has never been conclusively determined.
Meanwhile, in a fully autonomous system where AI agents interact with one another in complex ways, the potential for such events could increase. There is also the issue of systemic fragility. If many AI systems rely on similar data sources or training models, they may respond to market conditions in similar ways. This could lead to synchronised behaviour where multiple systems simultaneously withdraw liquidity or execute trades, creating sudden and severe market dislocations. In essence, diversity of thought, a hallmark of human-driven markets, may be reduced. Governance presents another challenge: when liquidity is managed by autonomous agents, who is responsible for their actions? If an AI-driven pool causes significant losses or disrupts a protocol, where does accountability lie? Developers? Investors? The AI itself? These questions do not yet have clear answers. Finally, there is the broader philosophical concern. As capital becomes increasingly autonomous, the role of humans in financial systems may diminish. Decision-making shifts from individuals and institutions to machines, raising questions about control, transparency and trust. At what point does the system become too complex for humans to understand? And if we cannot fully understand it, can we truly control it? All such questions are raised in Kenley Hamie’s, “The Agentic Commerce Manifesto” and begs the question is not, whether agentic commerce will arrive but whether London will still be a relevant player when it does?
The redefinition of capital
Liquidity has always been about access, the ability to move in and out of markets efficiently. But in a world of AI-generated capital, liquidity becomes something more - it becomes an active, intelligent force. Instead of waiting for human input, capital begins to seek out opportunities on its own; it flows where it is most needed, adapts to changing conditions and evolves. The boundaries between strategy and execution blur, as the same underlying intelligence handles both. This transformation has profound implications. Markets may become more efficient but also more complex. Opportunities may become harder to find but more precisely executed. Human participants may retain oversight but lose direct control. In this new paradigm, the most important liquidity providers may not be institutions or individuals, but systems, networks of algorithms that operate continuously, invisibly and autonomously. The rise of AI-generated capital challenges our traditional understanding of finance. It forces us to reconsider what it means to “own” capital, to “deploy” it, and to “manage” risk. Because when liquidity no longer requires humans, the question is no longer how markets function. It is who, or what, they are functioning for.
This article first appeared in Digital Bytes (7 th of April, 2026), a weekly newsletter by Jonny Fry of Team Blockchain.