Truth Before Execution.
Verisyn is the verification infrastructure for autonomous AI agents. Verify claims against trusted evidence before downstream agents execute.
Agents act on claims. Claims drift from the evidence that produced them.
Sources change
The spreadsheet, database, or API a claim was drawn from keeps updating. The claim, once written down, does not.
Errors compound downstream
An agent acting on a stale or fabricated number passes that error to every agent and decision after it.
Trust needs evidence
A confident-sounding claim and a verified one look identical until you check. Verisyn checks.
Every claim moves through the same fixed sequence.
One deterministic path from agent to verified output.
Claim extraction and evidence mapping are the only steps that use an LLM. Comparison, tolerance checking, and scoring are pure, reproducible functions - the same input always produces the same Trust Report.
Verisyn is MCP-first. The API is a second interface onto the same tools.
Runs the full pipeline, returns a Trust Report.
Detail on a single claim from a prior report.
Export a report as JSON or Markdown.
Service and configuration health check.
One endpoint. Full protocol support.
Verisyn's MCP server speaks Streamable HTTP at a single endpoint. Discovery is free; verify_claims is gated with an x402 payment challenge, identical whether called through MCP or the REST API directly.
-> initialize, tools/list, tools/call