Trust is not a promise. It is a record.
Every input, model request, tool call, question and output stays on the record — traceable, replayable, and owned by you.
Illustrative excerpt. A fictional session, with identifiers and decoded arguments shortened and models named by family. Each row comes from the table it names, in commit order, with the columns aligned for display; time is the row’s own timestamp: dispatched_at for a call, committed_at for a step and created_at for the rest. Routing, each model request and a tool’s own model call are separate invocations, so the record names the model behind every step — three in this one run. receipted means the result was recorded; a receipt can contain a tool error, and a retry is a separate call. Full results live in the referenced receipt objects. All ledger data is encrypted at rest and in transit.
Traceable.
“Why did the agent do that at 10:02 on March 14?” is one lookup — the tool call, its arguments and what came back. An answer, not an investigation.
Replayable.
Saved inputs, tool arguments, receipts and checkpoints preserve the evidence needed to inspect a run and evaluate changes against recorded work.
Versioned.
What was approved, what changed afterward, who changed it — every artifact answers all three.
Operator owned.
The ledger lives in your database — your schema, your keys, your region. Retention, redaction and hard-delete are your controls, not vendor tickets.
Exportable.
Plain SQL rows and artifacts you can query, dump and take with you. Compliance evidence is a query you run, not an export you request.
One context.
Every possibility.
Your team and AI, creating together — from first idea to finished work.