The accountability layer for AI agents.
Truss records mandate boundaries before execution and creates a verifiable chain of custody for every agent action. The difference between authorization and action becomes provable, never arguable.
The mandate gap
Your AI agent just executed a $500K purchase order. It was supposed to review the contract. Nobody told it not to sign. There is no record of what it was authorized to do, only a log of what it did.
This is not a hypothetical. AI agents are filing legal documents, approving transactions, and scanning codebases for vulnerabilities right now. The people who deployed them pressed go. The agents obeyed. And when something goes wrong, when a chatbot fabricates a bereavement fare policy, when a hiring tool discriminates against applicants over 40, when a model identifies software vulnerabilities it was never supposed to touch, the question is always the same: who authorized this?
Nobody can answer. Because nobody wrote it down.
Logs do not solve this. A log tells you the agent called an API at 3:42 PM. It does not tell you whether the agent was supposed to. A log tells you the agent signed a contract. It does not tell you whether signing was within scope. A log is a security camera. A mandate record is the key, the lock, and the authorization letter that says who gets in and what they are allowed to touch.
On June 12, this gap shut down the most capable AI model ever released. The US government wanted to know which agents were using Fable 5 to identify software vulnerabilities. It could not ask that question. No mandate records existed. So it did the only thing it could: it turned the model off for everyone on the planet.
What Truss does
Records mandate boundaries
Before an agent acts, Truss captures who authorized it, what it may and may not do, under which regulatory framework, and for how long. This is the power of attorney for AI agents.
Seals every action
When an action is taken, Truss creates a signed action record that links the action to the mandate. Ed25519 signatures, hash chains, tamper-evident. Cryptographic proof, not a dashboard.
Produces evidence packages
When a court, regulator, or counterparty asks whether an action was authorized, the answer is a record. Court-legible, tamper-evident, verifiable. Not an argument, a document.
Why it matters
The law is clear. California's AB 316 eliminates the “the AI did it” defense. Colorado's AI Act requires impact assessments and risk management programs. The EU AI Act mandates tamper-evident logging for high-risk systems. Singapore's Model AI Governance Framework for Agentic AI establishes that organizations remain legally accountable for their agents' behaviors.
Courts are already deciding. In Mobley v. Workday, the court held Workday liable as an “agent” even without a direct employment relationship. In OLG Hamm, a German court held a business fully liable for false statements by its AI chatbot. In the Amsterdam District Court, X Corp. was held directly responsible for harmful outputs generated by Grok.
The question courts ask is no longer whether an AI agent caused harm. It is whether the deployer can prove they exercised responsible control. Without a mandate record, the answer is no.
Built on TAP
The Tensflare Accountability Protocol (TAP) is the open specification that defines how mandate records, action records, and evidence packages work. It is designed to be implemented by any vendor, any framework, and any regulatory jurisdiction.
Built by Tensflare
Truss is the flagship infrastructure built by Tensflare Ltd. Founded in 2024, Tensflare was built by lawyers who spent years in private practice, regulatory roles, and capital markets, watching the same AI failures repeat.
Infrastructure must be neutral
Trust cannot be proprietary. If the accountability layer is owned by one vendor, it is not trust, it is dependency. We open-source our core standards because it is structurally required.
Empirical truth over narrative
We build on real data: 1,459+ hallucination incidents, $110K+ in sanctions, a government model shutdown. Numbers are our voice. Claims without mathematical proof do not ship.