For a long time, I was drawn to systems that remove the need to trust a single party.
That is part of what made crypto so important.
Bitcoin showed that money could move without relying on a central intermediary. Ethereum showed that applications could run on shared infrastructure without a company sitting in the middle. Those ideas mattered because they changed who gets control. They changed who gets to enforce the rules.
But over time I started noticing a deeper limitation.
The world is not only made of deterministic rules.
Most real agreements depend on interpretation. Most real disputes are not about whether a payment happened. They are about whether something was good enough, fair enough, complete enough, or true enough. Did the work match the brief. Did the supplier meet the standard. Did the other side act in good faith. Did the evidence actually support the claim.
Traditional blockchains are powerful, but they stop where judgment begins.
That started to feel like the real gap.
And now AI is making that gap impossible to ignore.
We are moving into a world where agents will increasingly transact, coordinate, negotiate, and act on behalf of people and companies. They will buy services, deliver work, evaluate outputs, and trigger payments. They will operate faster than humans and at much larger scale. But as soon as they do anything meaningful, they run into the same problem: who decides what happened when the answer is not binary?
That is the point where the old tools stop being enough.
Smart contracts are great when the world can be reduced to exact conditions. But a lot of economic life cannot. Human institutions handle this today through courts, arbitrators, platforms, and internal review processes. Those systems work slowly, expensively, and often opaquely. They were not built for a world of machine speed commerce.
The comforting assumption is that AI will somehow work around this. I think the opposite is true.
As execution gets cheaper, judgment becomes more valuable.
That is the shift I keep coming back to. For years, we focused on automating execution. Now the real bottleneck is decision-making under uncertainty. Not computation. Not storage. Not even coordination in the narrow sense. Judgment.
Once you see that, a different question appears.
What would it look like to build infrastructure for judgment itself?
Not a system that pretends ambiguity does not exist. A system that can process it.
That means moving away from the idea that every important outcome must come from one authority, whether that is a judge, a company, or even a single AI model. It means designing systems where decisions can emerge from plural verification, structured evidence, incentives, challenge mechanisms, and escalation. Not because humans are bad and models are good, or the reverse. But because any single decision-maker becomes a bottleneck and a point of failure.
The more I thought about it, the more it seemed obvious that this is where the next major layer of internet infrastructure has to emerge.
The internet has protocols for information. It has protocols for money. What it still lacks are protocols for credible judgment.
That is the problem I became obsessed with.
Not because it is academically interesting, but because I think it becomes economically unavoidable. If agents are going to transact at scale, there has to be some way to resolve ambiguity, verify non-deterministic outcomes, and enforce agreements that cannot be reduced to simple if-then logic.
Without that, we do not get autonomous commerce. We get automation trapped inside centralized platforms.
That distinction matters.
Because the real question is never just whether a system works. The real question is who controls the chokepoint. If every meaningful AI transaction ultimately depends on a centralized actor to interpret outcomes, then the value accrues there. The leverage accrues there. The market design accrues there.
So for me, this is not just about building better tools for crypto.
It is about building the trust infrastructure for a world where more and more of economic life is carried out by software. A world where agreements become dynamic, evidence becomes machine readable, and enforcement has to happen at the speed of computation without collapsing back into pure centralization.
That is why I care about this space.
And that is why I believe the next important breakthrough is not just trustless execution.
It is trustless decision-making.
That is the layer I want to help build.