Each generation of new technology changed the interface for market participants. At first, a person had to call a broker to execute a trade. Then came terminals like Bloomberg, comprehensive web platforms, and, finally, mobile apps, where opening and closing a position became a matter of a few taps. What united all of it was one constant: in every case, it was a human who looked at data on a screen and made a decision.
With AI now deeply integrated across the industry, that constant doesn’t work anymore, as the interface has just started to lose its central role. The system increasingly looks at the market first and makes the initial decision, while the human defines the boundaries and oversees the process.
That shift changes the structure of market participation itself.
The market changed first
There is a view that AI appeared from the outside and broke the established logic of financial markets. That view is wrong.
Markets had already outgrown the human operator long before AI became capable of replacing one. Today, serious market participation means following funding rates on multiple platforms, watching options positioning, tracking onchain flows, and absorbing macroeconomic releases simultaneously. A professional trader in 2015 could still process much of it manually, but by 2025, that has become barely realistic. The number of moving things at once has expanded faster than human attention could keep up.
In that environment, professionals had little choice but to delegate a large portion of the operational layer to automated systems just to keep functioning. Alerts replaced manual monitoring, a part of discretionary execution was substituted with algorithms, and the human moved from execution to oversight.
So it is wrong to say that AI suddenly removed traders from the market. The market itself had already made constant manual participation increasingly difficult, while AI is just accelerating a transition that was already underway.
Delegation without transparency is a risk
Most automated systems handle tasks without issues, so delegating them to AI is hardly a concern. Still, explaining what those systems did and why they made certain decisions is more difficult.
A system can be fast, accurate, and profitable, but remain opaque about how it reached a particular decision. In financial markets, that opacity can be punished severely. The user remains accountable for the outcome, whether or not they understand the process that was followed, and if the system gets things wrong, the cost is an immediate financial loss.
Once execution is delegated, performance alone is not enough, as the system also has to make its logic understandable to the user, to the risk manager reviewing it, and, when necessary, to the regulator asking questions after the fact.
In the Bank of England’s AI Consortium, participants described explainability and transparency as essential to trustworthy AI adoption in financial services, while also warning that inconsistent definitions can weaken risk management. The UK’s Financial Conduct Authority has likewise framed AI adoption around safety and responsibility in financial markets, arguing that balancing its benefits against risks is what makes this innovation both sustainable and enforceable.
This changes the status of explainability. It is starting to define the boundary between AI that can assist and AI that can be trusted to act.
Why does crypto reach autonomous execution first?
Assuming AI eventually gets explainability right and market participants can genuinely oversee the system, set boundaries, and define objectives and constraints, the next question is who adopts this model faster: traditional finance or crypto.
Traditional finance will unarguably get there, but crypto will move first. Legacy financial infrastructure was built around the assumption that a human authorizes every action. Settlement cycles, compliance workflows, and liability structures all reflect that assumption, since at each key decision point, the system was designed around direct human involvement. Crypto was not built this way.
Decentralized markets operate on programmable settlement, open protocols, and interoperable infrastructure that do not depend on a person being present at every stage of the transaction cycle. As a result, autonomous systems do not need to be retrofitted, but are already compatible with it by design.
In traditional finance, agentic execution requires regulatory alignment, institutional redesign, and changes to control frameworks that were built for a different operating model. In crypto, many of those constraints are lighter, which makes it a more immediate testing ground for systems that can interpret intent, retrieve data, evaluate choices, and act within predefined limits.
For instance, earlier this year, a Layer-1 blockchain introduced infrastructure that allows AI agents to execute onchain derivatives workflows through natural language, bringing autonomous execution directly into live markets. Traditional finance has no true equivalent yet, because the underlying architecture still makes this model harder to deploy at scale.
To be fair, that gap will narrow over time. Traditional financial institutions will eventually adopt similar models once questions around liability, compliance, and model accountability become clearer. But by the time they do, crypto will likely have already accumulated operational experience. It will have learned what works, what breaks, where trust fails, and how intelligent execution can be made more reliable.
The next competitive layer in finance
Taken together, the competitive field in financial markets is changing. It used to be about the interface, the UX, the number of tools available, or even the AI capability itself. Now it comes down to the quality of delegation, which means how accurately a system interprets intent, how reliably it operates within defined limits, and how clearly it accounts for the actions it takes.
From where I stand, crypto is likely to produce the first meaningful answers to those questions. The infrastructure is already compatible, and firms have already begun testing live models, so what emerges there around trust, explainability, and autonomous execution will help shape the next operating standard for financial markets.
