What happens when an AI agent takes an action you can't prove it took - or didn't take? That's the compliance nightmare Viatoris was built to solve.
The tool, launched by a solo developer, addresses a gap that most enterprise AI deployments quietly ignore: standard logs are mutable. Anyone with database access can alter them. When a regulator or customer asks "what did your agent do and when?", a log file is not proof.
Viatoris gives each AI agent a cryptographic identity using W3C DID (a standardized format for digital identities, similar to a verifiable passport for software) paired with an Ed25519 key (a widely-used cryptographic signing algorithm). Every action the agent takes produces a signed receipt. Alter that receipt after the fact and the cryptographic signature breaks, making the tampering detectable. Anyone can verify an agent's identity and action history through a public API.
The practical use case is straightforward: a financial services firm deploys an AI agent to process customer requests. Six months later, a compliance audit asks for proof of exactly what the agent approved, when, and under what conditions. Without something like Viatoris, the answer is "here are some logs" - which isn't the same as proof.
Who This Is Actually For
This isn't a tool for someone running a ChatGPT wrapper for their blog. The audience is developers and IT teams deploying autonomous agents in regulated industries - finance, healthcare, legal, insurance - where the question of agent accountability isn't theoretical.
The concept borrows from how software supply chains handle provenance (tracking where code came from and whether it was tampered with), applying the same principle to agent actions rather than code artifacts.
Viatoris is early-stage, built by one developer who kept hitting the same wall in enterprise deployments. The public API for verification is a smart design choice - it means any third party can independently confirm an agent's history without relying on the deploying company's own records. That independence is exactly what compliance teams need.
The broader signal here: as AI agents move from demo projects into production workflows that touch money, health data, and legal decisions, the tools that handle accountability and auditability will matter more than the agents themselves.