An audit to find the work worth automating, and a monthly fee to run what we build. Agent development is quoted per workflow from the audit, so you approve each one on its own numbers.
The audit prices each build separately against the hours it returns. Nothing is bundled, so you can take the three that pay for themselves and leave the rest.
Ten working days inside your operations. Workflow map, hours ledger, costed build plan, and an access review, all yours whether or not you build with us.
Per workflow. Prompts, eval suite, connectors, the handoff rule, and a rehearsed rollback, priced from the audit rather than guessed at.
Per month, whatever the number of agents. Monitoring, eval upkeep, cost tuning, same-day fixes on broken runs, and the monthly hours ledger.
All thirty-eight, plus new ones your workflows require. A connector we build for you stays in the library for everyone, which is why we do not charge for it.
Most clients never pass it. Above that we pass through at cost, itemised per workflow in the monthly ledger, with no markup.
Two sessions with your engineers and the documentation to go with it. Some clients take managed operations in-house after a year, and that is a fine outcome.
Two refunds in five years. Both teams had already automated the obvious work themselves, which is exactly the answer we would want to give them.