Process mining reads clicks and event logs, it sees what systems recorded, not why a decision was made. Discovery interviews the people who do the work, so it captures the judgment, the exceptions, and the rules that never get keyed in. No logs, no install.

No. Discovery starts with interviews, not integrations, and nothing is installed to begin with. System access is earned later, only when production data is actually needed, so nothing waits on a data project.

One department, mapped. Every process ranked and priced by volume, cycle time and automation fit, each figure marked observed, estimated or assumed, plus a build-ready runbook and the cases to test the top one. The description is yours either way.

The people who actually touch the work, a handful of short interviews across the department. No steering committee and no six-month workshop series. Ten minutes each is often enough.

No. Zero retention, not by us and not by our model providers. The same clause holds all the way down the chain, and it lives in the contract, not a blog post.

Everything that matters. Agents prefill and propose, your people approve. Review points are defined during discovery, not bolted on later, and high-stakes steps stay behind human approval.

That disagreement is the signal, not noise. Discovery marks a claim disputed, holds it open, and keeps it there until more sources confirm it. Nothing is averaged into a tidier answer.