Everything, answered plainly.
If your question is missing, send it — answers land here within the week.
Operations, security, and honesty.
Connect two tools and it runs the four flows from the homepage — refunds, digests, incident write-ups, receivables. Most teams add their own flows in week one.
Only where you allow it. Every flow ships with an approval gate; you loosen them per task, per amount, per tool — and the log records every change.
It stops and asks, with its plan and evidence attached. Uncertainty is a first-class state, not an error.
Every run is logged step by step — inputs, tool calls, outputs — and replayable read-only. Auditors stop asking for screenshots.
OAuth into your tools, import your policies, run your first task from the console. Under an hour, no engineers required.
Runs are reversible by design — drafts held, sends spaced, refunds limited. The log shows precisely what to undo, and the policy that let it happen gets a review prompt.
A mix, chosen per step and benchmarked monthly. You can pin a provider on Scale. Model choice never changes the verification loop.
Your region, your retention policy. EU residency ships on Team and Scale. Credentials are scoped, encrypted, and never used outside a logged run.
Yes — once you open the gates. Teams typically reach hands-off on refunds and digests by week six, with the audit log as the confidence meter.
Export runs, logs, and artifacts any time — JSON or SIEM stream. Leaving should be as clean as staying.