What we are
founded on.
The founding conviction
Most organizations employ knowers. Almost none are knowing.
The difference is not the quality of the people or the ambition of the work. It is whether the reasoning behind decisions — the sources, the warrant, the judgment chain — is preserved in a form that the next person can inspect, challenge, and build upon. Without that, even the best thinking disappears when the session ends. What remains is the conclusion, stripped of everything that made it credible.
This is not a new problem. It is why common law encodes its reasoning in written opinion, not just verdict. Why the best research frameworks outlast the researchers who built them. Why some institutions accumulate knowledge across generations and others simply accumulate tenure.
What is new is the scale at which the problem is now arriving. AI augments human reasoning in ways that are genuinely powerful. But almost nothing being built governs what that reasoning produces — who warranted it, what it depended on, whether it can be trusted by the next person who needs to act on it. The volume of AI-augmented judgment is growing faster than any institution’s capacity to make it inheritable.
That is what Tsari is built for.
What we believe
The unit of value is never the output. It is the warrant.
A decision, a finding, a ratified claim — these are not valuable because they exist. They are valuable because their reasoning can be traced, challenged, and built upon. Outputs without warrant are not intelligence. They are performance.
Provenance is not a feature. It is the definition of knowing.
A claim without a traceable source is not a weak claim — it is not yet a claim. Provenance is not something added to intelligence to make it trustworthy. It is what distinguishes intelligence from noise in the first place.
The audience is always the future inheritor.
Every record, every ratified decision, every piece of governed reasoning must be legible to someone who was not present when it was created. We design for the person who inherits the work, not only for the person who does it.
Speed is not the problem. Unwarranted speed is.
Tsari is not against fast reasoning or AI-augmented work. The constraint is that nothing should inherit authority it has not earned. Ratification is the gate, not the bottleneck.
We must be what we build.
Tsari is itself a knowing institution or it is nothing. The doctrine files, the decision trails, the provenance statements — these are not administration. They are the proof.
What we are building toward
A world in which the reasoning behind consequential decisions — in research, in institutions, in AI-augmented work of every kind — is as durable and inspectable as the decisions themselves. Where the gap between what AI produces and what institutions can trust, trace, and act on is not a permanent condition but a solved infrastructure problem.
We are early. The infrastructure does not yet exist at the scale it needs to. Building it is the work.
Tsari is derived from Hausa — the language spoken by 94 million people, primary in northern Nigeria and Niger. In Hausa, tsari means: order, system, structure, method, schema — and also shelter, refuge, protection. The institution is named after what it provides.