Tuent is a startup built out of Silicon Valley, based in Santa Clara, California. Two founders who have built together before, now on the problem of what AI agents do once they are inside a company. The company exists on one conviction: you cannot govern an agent with something it can talk to. If you ever stop using Interpose, the report, the policy, and the trail stay yours.
Runs the pilots, the partnerships, and the company. Has done research in agent security, contributed to several frameworks in the field, and speaks regularly on the current and coming need for security that covers the whole agent, not just its prompts.
Founding engineer and the company's principal thinker. Designed the gate, wrote the enforcement engine and the test suite behind it, and works the problem as a philosopher first: what an agent intends, and what a security team can defensibly claim from the record. Installs Interpose with the customer's team in every pilot.
One conviction, held since the first line of the enforcement engine: you cannot govern an agent with something it can talk to.
Most AI security tools use AI, so a prompt that steers the agent can steer the guard. Interpose is deterministic end to end. The same input gets the same answer every time, there is no model in the decision path, and that is what makes the control auditable rather than persuadable.
Every decision Interpose makes is hash chained and Ed25519 signed, so the record survives the agent, the developer, and us. Agent autonomy is doubling every few months, and per action human review is already arithmetic nobody can afford. The layer that governs agents has to be the one thing in the stack that does not think.
Security, cloud partnerships, network visibility, and AI research. They review the architecture and the pricing. Named on request.
Due diligence questions, a security questionnaire, or an intro: [email protected] or [email protected]. A founder answers, usually the same day.