We kept shipping agents and keeping no record.

Priya leads engineering at Assembly. She spent a decade building distributed tracing for backend systems before noticing that the fastest-growing category of software in the enterprise — agents acting on a person’s behalf — had no tracing at all.

Tomas builds the instrumentation layer — the part that captures what an agent was asked to do, what it touched, and what came back.

Mei-Ling shapes what Assembly builds. She joined from an enterprise AI programme where she was accountable for adoption numbers she could not defend.

Krishna leads the team behind Assembly’s verification layer — the part that reconstructs an agent’s run step by step, after the fact.

Oliver is responsible for how Assembly handles what it records — scoping, retention, and the parts auditors ask about.

Yuki designs how Assembly presents evidence — the difference between a number a team acts on and one they learn to ignore.

Naomi runs Assembly Research and edits the Assembly Index, the benchmark of agent reliability drawn from anonymised invocation data across the customer base.

Daniel built Assembly’s correction detection — the models that spot when a human quietly rewrites what an agent produced, without ever filing feedback.

Hannah studies what happens in the minutes after an agent completes a task — the edits, the re-runs, the quiet abandonment.

Aisha leads the team that deploys Assembly inside large organisations — which means she has catalogued more undocumented agents than anyone should have to.

Rafael designs Assembly deployments for regulated industries, where observing an agent means proving what data it touched.

Sofia works with Assembly customers after go-live, on the harder half of the problem: who owns each agent, what gets reviewed, and when one should be retired.