The question was simple.
The answer wasn't.

What was our churn last month? Lost logos, says Sales. Lower MRR, says Finance. MAU decline, says Product. And what do you mean "month"? Calendar? Last 30 days? What sounded like a simple question becomes a negotiation. Days later. Sometimes weeks.

And this is just one metric. The chief of staff at the CFO office of one of our customers used to spend more than 30% of his time on data reconciliation, not because he was inefficient, but because he knew who to ask. He was the system. The human API bridging SAP, Salesforce, Workday, Netsuite, and a dozen other sources, each with its own definition of every metric that matters.

Now imagine he's gone, and AI is next in the queue.

For decades, enterprise software captured records. What it never captured was meaning. Every system learned its own language. Every department developed its own dialect. As long as humans were in the loop, people who could translate, reconcile, and make sense across systems, the work got done. Slowly, expensively, but done.

Then came AI. Powerful in isolation. But AI doesn't fix what's underneath. It inherits and amplifies the ambiguities. An agent querying five systems with five definitions of revenue doesn't produce insight. It produces confident noise. Faster than before. And it's the number one reason AI rollouts are stumbling.

Ohno built the Toyota Production System to eliminate the waste hiding inside every manufacturing process. We're doing the same for enterprise knowledge, eliminating the reconciliation work that is wearing smart people out, acting as human APIs between systems, and blocking AI from delivering on its promise.