Discovery interviews: requirements gathered by conversation

Discovery interviews

Requirements gathered by a conversation, not a form

Kickoff questionnaires get put off. A ten-minute conversation with a character that asks one question at a time, follows up, and reads back what it heard does not. WorkPM runs discovery through Workup Agent, and the answers arrive in the project before the first call.

1

You write the questions

Once, in plain language, with an optional hint for what a complete answer looks like. Eight to twelve per session works best.

2

The client gets a link

A two-line email from you with a private link. It can be opened on a phone, paused, and resumed. A reminder goes out once if it sits untouched.

3

Workup Agent asks

By voice or text, one question at a time. It asks one follow-up when an answer is thin, accepts "I need to check with someone", and never answers on the client's behalf.

4

You get the sheet

Each question, the answer, the client's own words, and what is still open, in the project and in your inbox, with the transcript and a CSV.

What makes it trustworthy

  • Every recorded answer carries the client's exact wording and the question it answered. Nothing is paraphrased into something more specific than what was said.
  • Skipped questions stay open and visible, so the kickoff call starts with the gaps, not a guess.
  • At the end, the client hears the recap and can correct anything before it is sent.
  • The same engine that runs your client-facing intake runs discovery, so the client meets the product on their own project.

Good questions to start with

  • Which systems hold the data today, and which one is the source of truth?
  • Who signs off on scope changes, and how do they prefer to be reached?
  • What does "done" look like for the person who will use this every day?
  • What went wrong the last time a project like this was attempted?