Zeeshan Khan

The Verification Series

Generative models pushed the cost of producing a plausible claim to near zero. The cost of checking one didn't move. That gap is the subject of these seven essays.

The argument runs through them in order. Verification, not generation, is now the constraint. An agent that acts on your behalf needs a record a counterparty can check — and that record has to survive the network being unreachable. A signature proves a document is intact, not that it's honest, which is the difference between a claim and a credential. Detecting that something broke is not the same as knowing what the break touched. And the same problem arrives at the top of an engineering organization, where a leader who has stepped fully back is grading work through a report the tools have learned to make convincing.

One idea underneath all of them: a maker cannot grade its own work. As generation gets cheap, value moves to whoever can still independently verify. That's true of agents, of supply chains, of credentials, and of the people running engineering teams.

Written May through August 2026. Best read in order, though each stands alone.

AI Made Generation Cheap. Verification Is the Bottleneck Now.

Generative models pushed the cost of producing plausible content to near zero. The cost of checking whether it's true didn't move. Why verification, not generation, is the constraint that now shapes everything downstream.

~8 min read · May 2026

The Agent Passport

If agents are going to act for us, they need to carry something a counterparty can check: what they are, what they're authorized to do, who answers for them. On what that record must contain — and the one question it can't answer about itself.

~10 min read · May 2026

When You Can't Phone Home

The verification machinery being built for agents assumes a network is always reachable. On what trust looks like at the disconnected edge — carrying the authority, reconciling the record, choosing the failure posture — and the one property, revocation, that no offline system can fully deliver.

~6 min read · June 2026

A Claim Is Not a Credential

A passport's value lies not in what it says but in who signed it. On the difference between a self-signed assertion and an independently issued, revocable credential — and the three things, independent issuance, revocability, and recourse, that only the second has.

~6 min read · July 2026

An Alarm Is Not a Map

A red checkmark tells you something broke. It doesn't tell you what the break touched, who depends on it now, or what to do about it. On the distance between detecting tampering and understanding damage — and why that distance is the entire frontier of integrity tooling.

~11 min read · July 2026

Leading AI Is a Trust Discipline

The tools now produce work that looks trustworthy without being trustworthy — and fool the maker most of all. Why a leader's judgment decays at a distance, and what staying close to the work actually means.

~6 min read · August 2026

A Maker Cannot Grade Itself

The bold engineer, the self-verifying machine, and the junior you didn't hire are the same problem: a maker can't grade its own work. On a leader's real job — being the grader the work can't fake, and raising the ones who'll do it next.

~7 min read · August 2026

The companion series, Forward Deployment, asks a related question: where the person who can verify has to be standing.