Markets Used to Be Places
For most of modern history, markets were treated as bounded places. They were venues that professionals visited, institutions intermediated, and ordinary people touched only occasionally. Finance was a vertical. Markets were where price discovery happened, but they still felt separate from the rest of life.
Crypto widened that frame. It did not just create new assets. It created new ways to express conviction, speculate, hedge, coordinate, and surface information through open, programmable systems. DeFi turned financial primitives into software. Prediction-market infrastructure is now credible enough that major news organizations are feeding data into it. Culture, politics, memes, and macro increasingly started to flow into the same machine.
Once you see that shift clearly, the next one becomes easier to understand. Markets are no longer just places people visit. They are becoming infrastructure other systems can call.
Mass Participation Changed the Surface
One of the most important changes of the last cycle was not technical. It was cultural. Markets became mass-participation environments. That does not mean everyone became a disciplined investor. It means that more people began using markets to express conviction, seek upside, form identity, and react to events in real time.
The result is that finance increasingly bleeds into culture, and culture increasingly bleeds back into finance. A trend, an event, a political outcome, a product launch, or a social narrative can all become something a market prices. Permissionless systems make that possible because they lower the cost of creating new markets and of participating in them before institutions have fully decided they belong. The January 2026 prediction-market literature is already formalizing these markets as noisy but information-bearing inference systems rather than curiosities. Recent price-discovery work on retail herding says something similar from another angle: information quality itself changes once participation patterns change.
This matters because it means the object called “finance” is already much broader than a bank account, an options chain, or a Bloomberg terminal. It is becoming one of the main ways the internet expresses and updates shared economic beliefs.
Markets Are Better Than Ordinary APIs
A normal API returns stored data. A market is different. A market produces information through adversarial competition among participants willing to risk capital on their beliefs. That makes market outputs unusually expressive. They are not just records of what has happened. They are live signals about what participants believe is likely, mispriced, or changing. That still does not make them neutral truth: the latest sociotechnical audits of prediction markets explicitly warn that markets also compress noise, asymmetry, and strategic behavior into the signal.
Onchain markets push this further because they are easier to compose into software. They are programmable, permissionless, globally accessible, and increasingly cheap to query and settle against. That means prices, yields, spreads, probabilities, and flows start to behave like machine-readable endpoints that other systems can consume directly instead of outputs meant only for humans staring at charts.
Once that happens, a market stops being just a destination for speculation. It becomes a general-purpose decision surface.
Agents Will Consume Markets Differently
Humans use markets intermittently. We read them, interpret them, and decide what to do next. Agents can consume them continuously. They can query prices, probability surfaces, funding rates, liquidity conditions, and venue state directly and fold those signals into their own workflows in real time. Broker and venue APIs are increasingly making that style of direct machine consumption a normal product surface.
This is why it makes sense to think of markets as APIs. An agent does not need a market only as a place to trade. It may need a market as an input into research, execution, routing, monitoring, underwriting, or budget allocation. The more financial systems become programmable, the more natural it becomes for software to treat markets as information infrastructure.
That shift is easy to miss if you only look at markets through the old lens of “what humans buy and sell there.” The more important question is “what other systems can do once they can read and act on these market outputs directly?”
Finance Becomes a Horizontal Substrate
The deeper implication is that finance stops looking like a distinct software category and starts looking like a horizontal substrate. Payment rails, stablecoins, lending markets, prediction markets, and programmable exchanges become pieces other products can build on rather than isolated destinations with their own walls. Mainstream brokers are now explicitly experimenting with tokenized-asset rails that fit that thesis.
This is also why the old distinction between “financial software” and “everything else” gets weaker over time. If markets increasingly generate live, costly-to-fake information that software and agents can consume, then finance becomes part of the fabric of how other products reason, route, settle, and decide. That does not make every app a trading app. It means that more products will quietly rely on financial signals under the hood.
Why Financial Markets Become the First Interface
If markets are becoming programmable infrastructure, then financial markets become the first natural interface between humans, agents, and the wider machine economy. They are already dense with live signals, explicit feedback, changing probabilities, visible risk, and measurable outcomes. That makes them a much better proving ground than slower, fuzzier categories of software where the consequences are delayed and the system can hide whether it is actually working.
This is why trading comes first. Not because trading is the final product, and not because every user ultimately wants to become a trader, but because markets are the place where software already has to read signals continuously, form intent under uncertainty, route across fragmented infrastructure, and act with discipline. If you can build an interface that helps a human delegate that loop safely in markets, you are much closer to building a real financial operating layer everywhere else.
In that sense, financial markets are not only an application domain. They are the first operational surface where the future of agentic finance becomes visible. They are where the abstractions break fastest, where trust has to be earned, and where the control layer has to prove it can do more than narrate. That is exactly why they become the first interface.
What This Means for General Liquidity
This is one of the main reasons General Liquidity exists. If markets are becoming programmable infrastructure, then someone needs to build the control layer that makes them usable for real operators and, eventually, for the humans who delegate work to software. Raw access is not enough. Context, trust, memory, approvals, routing, and judgment still matter.
Gordon starts with trading because markets are the hardest proving ground for this kind of system. But the longer-term direction is broader. An agentic financial terminal only makes sense if you believe finance is becoming a layer of programmable infrastructure rather than a silo. That is exactly the transition now underway.
If everything becomes software, and more of software starts consuming market structure directly, then the next important question is not whether everything is finance. It is how much of the world will start depending on markets without calling them that.