How are you producing verifiable evidence of what your agent actually did — not logs, not telemetry, but something a third party could independently verify after the fact?

Working on an open standard for this (ACTIS) and genuinely unsure if we're solving the right problem. What does your current audit trail actually look like in production?