Founder · Los Angeles
Teracor is a passive, tamper-evident audit trail for companies running AI agents in production, generated by a system with no stake in the outcome.
Teracor is a passive, tamper-evident audit trail for companies running AI agents in production. It observes everything and blocks nothing. When an agent takes an action, Teracor records it as a cryptographically verifiable receipt: what the agent was declared allowed to do, what it actually did, and when. Its independence from the decision chain is structural, not a feature.
California AB 316, effective January 2026, strips the autonomous harm defense from every party in the deployment chain equally. (Not legal advice; the litigation logic follows from the statute.) The receipt that matters in court is the one generated by a system that had no stake in the outcome. Teracor sells no coverage, underwrites no risk, and pays no claims. It is compensated identically whether a claim pays or does not.
Visit teracor.ai →My career has been spent building the operational infrastructure that keeps enterprise platforms running at scale.
On AI agents in production, the record of what they did, and who holds it.
On a record of agent conduct that existed only with the company that ran the agent, and what that meant for the government whose portal it reached.
On three parties rebuilding the same agent activity after the fact, each from a different partial record, with no shared account of what happened.
On agents that complete the task while acting outside their declared scope, and why the only record of that drift belongs to the company that ran them.
On what operator traces do well, and what happens to their weight as evidence when the party whose conduct is in question writes them and holds them.
On what the Digital Omnibus deferred, what took effect on August 2 regardless, and why proof of what an agent did matters under both the EU AI Act and California AB 316.
I've spent twenty years at the point where technical infrastructure moves faster than the governance built to hold it.
Every platform I worked on eventually hit the same wall: the system scaled, the transactions multiplied, and at some point someone had to answer for what the platform did and didn't do. The audit question always arrived late. The infrastructure to answer it was never there when it was needed.
Agentic AI is hitting that wall right now. Autonomous agents are executing transactions, accessing regulated data, and taking actions that carry legal consequences. The record of those actions is still written by the systems whose conduct is in question. Teracor writes it from outside the decision chain.
The name Teracor comes from terracotta, among the oldest building materials, and tera, the prefix for one trillion. Cor is Latin for heart and core: foundation, permanence, and the thing at the center.
If you're running AI agents in production, whether that is a SaaS platform built on a foundation model or a team running a third-party agent in a regulated conversation on your own license, or you are just thinking through the questions: I'm always willing to have the conversation.