YC's Company Brain: Extraction Is Necessary. It Is Not Sufficient.

YC just published a request for what they are calling the Company Brain. Tom Blomfield described the problem well.

Every company's critical knowledge is scattered. It lives in people's heads. In old Slack threads nobody reads. In tribal knowledge that walks out the door when someone resigns. AI cannot operate reliably without it.

So the bottleneck to AI automation is no longer just the models. It is the missing organizational context.

They are right about the problem.

The second half of the problem

There is a second half of the problem that extraction alone will not solve.

The obvious starting point is extraction. Mine the Slack threads. Parse the old emails. Pull the support tickets. Build a structured knowledge file. Give it to the AI so it can act.

That will help. But it has a limit.

The knowledge that matters most was often never written down anywhere.

• How does your procurement lead handle a vendor who has preferred status but missed an SLA?
• Which escalation paths does your ops team bypass in practice—and why?
• What does your best invoice reviewer catch that the policy document does not cover?

That judgment does not usually live in a document. It appears at the moment your team makes a decision. So you cannot only mine it from the past.

You have to capture it going forward, one decision at a time.

Not just extraction. Accumulation.

That is the architecture we have been building at Wayvo.

Not just extraction. Accumulation.

• Every time your team approves an output, the agent learns: this is what good looks like here.
• Every time they edit one, the agent learns: in this context, this organization does it differently.
• Every time they override one, the agent learns: the standard approach does not apply here.

No action executes without human sign-off. That is not just a safety constraint.

It is the mechanism by which the system gets smarter. The sign-off is the training event.

Looking backward and forward

A lot of Company Brain approaches will begin by looking backward. That is absolutely necessary. But it is not sufficient.

The organizations that win will be the ones whose AI systems also capture judgment forward—from the decisions their teams make every day.

The knowledge that matters most was often never written down anywhere.

The only way to get it is at the moment of decision.

Where does the most valuable institutional knowledge live in your organization—and what happens to it when your best people leave?