FDEForce
Sign in
Docs

How FDE Force works

You run FDE Force by talking to your AI agent: Claude Code, Codex or another. FDE Force holds your workspace, your customers and what their people tell the interviewer; your agent does the thinking. These are the steps, from setup to follow-ups.

  1. Get your login

    Ask us for an FDE Force login: email [email protected] with your company name. You get a username and password for your own workspace. Only your company can see it.

  2. Connect your AI agent

    FDE Force works with Claude Code, Codex or any agent that speaks MCP. Add it to the one you use.

    claude mcp add --transport http fde-force https://getfdeforce.com/mcp

    Then run claude, type /mcp, choose fde-force and pick Authenticate.

    A browser page opens. Sign in with your username and password, then click Allow. Your agent only ever sees your own workspace.

  3. Set up your workspace

    Your workspace describes your product once, for every customer. Ask your agent to set it up. It reads your website and docs, asks you for what it can't find, and saves:

    1. What your product is, who uses it, and what it does not do.
    2. Your features: what each one does and what it needs to be set up.
    3. Your logo, shown to the people you interview.
    4. Your roles: the users and buyer personas you meet at a customer. Each role has a description and six general questions, the template every customer starts from. One of them is always the general role, which every interview starts on.

    Say to your agentSet up our FDE Force workspace from our website and docs. Ask me for anything you can't find.

    Settings in the console shows the setup progress, with a tick for each part that's done.

  4. Add a customer

    Each customer you deploy into gets its own details. Your agent asks you for them:

    1. Customer profile: what you know for certain. Who they are, their size and sites, the systems they run, why they bought.
    2. Business goal for this deployment: what the customer wants the rollout to achieve, agreed at the outset, and how and by when they'll judge it.
    3. Learning goals: what you need to learn there, most important first, and the roles each one is for.

    Then your agent copies every role's questions from your workspace onto the customer and adjusts them for this customer, working in the learning goals. It shows you what it changed. These are the only questions the interviews ask, and you can see and edit them by role in the customer's Customer details tab.

    Say to your agentAdd Brook & Co as a new customer. Here are my notes from the sales calls.

    Change any of it in the Customer details tab. Every interview from then on uses it. Changing your workspace later doesn't change a customer's questions.

  5. Run the first interviews

    Each customer has one link. Send it to the people whose work your product will touch, from the sponsor to the people doing the work. Your agent can draft the message.

    1. They open the link, enter their name and work email, and talk to the interviewer by voice in the browser. It takes about ten to fifteen minutes.
    2. Every interview starts on the general role and asks what they do. Once it knows, it switches to that role's questions for this customer.
    3. It works from your product facts and the customer's details, asks one open question at a time, and ends by asking who else to speak to.

    Say to your agentDraft a message to send Brook & Co's head of audit with the link.

  6. Roll up each interview

    After an interview, your agent folds it into the customer's deployment map. FDE Force has no AI of its own, so this is work it hands to your agent.

    1. The deployment map: how the customer works, where your product fits and what to set up first, ranked against their business goal, plus open questions, disagreements and who to talk to next.
    2. What each role has told us: short, anonymous findings that the next interviews with that role test.
    3. Questions for you: anything people asked that the interviewer couldn't answer.
    4. Learning goals: how far each one is covered, and what is still open.

    If something in the map is wrong, or someone said something in confidence that shouldn't be repeated, ask your agent to correct it: the map, a conversation's summary, or what a role has told us. Every map correction is kept in its history.

    Say to your agentCheck FDE Force for open work.

    The customer's Overview and Deployment map in the console show the result.

  7. When people come back

    People sign back in with the same email, on any browser. What they get depends on where they left off:

    1. If the call dropped or the browser closed, they pick up where they left off.
    2. If you have assigned them a follow-up, they get that (next step).
    3. If they have finished their interview, they can add to it. The interviewer has the earlier conversation and asks about what wasn't covered.

    Adding to an interview never counts as a second person in the map.

  8. Follow up with one person

    A follow-up is an interview written for one person, about one subject. Use it to go back to someone with questions the map still has open, or to brief an interview for someone a colleague named.

    1. Ask your agent for one. It drafts it from what the map is missing for that person: open questions for their role, disagreements they could settle, why they were suggested.
    2. Go back and forth on it: what it's about, up to six questions, what we already know so it isn't asked again, what the interviewer may share, and what to stay off.
    3. When you're happy, your agent assigns it to their email and drafts the message to send. They see it the next time they open the link.
    4. Only their answers to the follow-up's questions go into the map.

    Say to your agentDraft a follow-up with Carl at Brook & Co about how audit files get signed off.

    The customer's Follow-ups tab shows each one and where it's at.