An AI agent is not a single call. It is a sequence of choices: read this, route to that model, call this tool, remember that fact, spend here, hand off there. Each choice changes what happens next. When something goes wrong — or when someone asks why the agent did what it did — the honest answer is usually a reconstructed story from logs the operator controls. That is a claim, not proof.
AI agent receipts close that gap by attesting the agent's actions as they happen.
What an agent receipt captures
An agent receipt binds the conditions of a single agent action into a signed record. Across a run, those records form a chain — a verifiable timeline of what the agent actually did:
- Route decisions — which path or model the agent selected at a branch point, and against what inputs.
- Tool calls — which tool was invoked, with what arguments, at what step.
- Model choices — which model handled which segment, so a downgrade or substitution can't happen silently.
- Memory and state — what the agent carried forward, so provenance survives across steps.
Each is signed with post-quantum ML-DSA-65 and verifiable offline by anyone with the public key. The receipt attests what happened, not whether the decision was optimal — a narrow, honest claim that is exactly what makes it defensible later.
Why a chain, not a single stamp
Agents are judged on the whole trip, not one hop. A single signature over a final output tells you nothing about the seven decisions that produced it. A receipt chain lets a reviewer walk the sequence — decision by decision — and confirm none of it was altered or backfilled after the fact. Providers own the roads the agent travels; the agent owns the trip; the receipt is the record of the route.
See it run
The clearest way to understand agent receipts is to watch one build. Agent Trip streams a live sequence of route decisions, minting and verifying a receipt at each step. It is the agent-actions surface of the same receipt primitive that signs inference and human-origin data.
How much proof does the task need?
Not every agent action needs the same weight of assurance. A low-stakes lookup and a regulated financial handoff sit at different points on the spectrum. Hive's AI assurance tiers map the same receipt primitive across levels — from signed inference to verified agents to regulated workflows — so you choose the amount of independent, third-party proof a given task actually requires. Verification is always free; you only pay to sign.
Watch route decisions get signed and verified in real time, then see how assurance tiers scale the proof to the task.