We scanned Documenso. Here is its journey map.

26 routes. 14 components. 24 customer-facing copy strings. 6 coverage gaps. And seven workflows, placed on the customer journey: Sign Up, Sign In, Send Document for Signing, Sign a Document as a recipient, Create Template, Manage Organisation Billing, and Invite Team Members.

That last part matters. The scan does not just list screens. It arranges them into the journeys a real user walks, stage by stage, from first visit to paid plan.

1. The empty documents page.

A new user lands on the documents list with nothing in it. Documenso's own empty state says "We're all empty". The scan flagged the moment from the code and named the risk: "User doesn't know how to start — empty state paralysis before first upload". The copilot it designed for that moment says: "Drag a PDF here or click to browse. You can add multiple documents to one envelope."

2. Field placement.

Sending a document means placing signature fields on a PDF. The editor is powerful and quiet, and the scan marked this as the workflow's biggest stall risk. The intervention it proposed is a voice walkthrough: "Let me show you — drag a Signature field from the sidebar onto the spot where the signer should sign. You can resize and reposition it afterwards."

3. The failed submit.

A recipient fills in every field, clicks Complete, and the request fails. The scan anchored an intervention to the exact error toast: "We were unable to submit this document at this time". Its answer is written for that moment: "Something went wrong — please try again. Your progress is saved, so you won't lose any filled fields."

When a returning user pauses on the sign-in form, the scan chose silence as the intervention. Its reasoning, verbatim: "Sign-in is a low-friction moment — the 'Forgot your password?' link is already visible. No intervention needed." The goal it recorded: "Avoid interrupting a routine action."

Help that knows when to stay quiet is the whole point. A tour would have started talking.

The scan reads code, not production traffic. Its risks are hypotheses a founder would recognize, not measurements. It sees the copy users see, but only what lives in the repository. And very large monorepos take longer than this one did. The map is a starting staff briefing, not a crystal ball. Rehearsal against simulated users is where the hypotheses get tested.