ZeenTech Interviews

How to use

Five steps, about 15 minutes

ZeenTech Interviews turns your AI assistant into an interviewer. It asks one question at a time, saves your answers, and shares them only with the people you choose.

1. Sign in

Open the link in your invite email, type your name, and press Email me a code. Type the 6-digit code from your inbox. There is no password. Next time, sign in at interviews.zeentech.qa/login with a fresh code.

2. Connect your assistant

Go to Agents and choose how much each assistant may do:

Claude Code

Press Add agent. You'll get a passphrase and a ready-made command; paste it into a terminal. The passphrase is shown once, so keep it safe like a password.

Claude (web or desktop) or ChatGPT

In your assistant's settings, add a custom connector with the address https://interviews.zeentech.qa/mcp. A ZeenTech page opens: sign in, choose what the assistant may do, and press Allow.

3. Do your welcome interview

Say to your assistant: "Run my welcome interview." It asks about 11 short questions, one at a time, and saves your answers. If you're unsure about something, say so; it's marked as a draft. Say "skip" for anything you'd rather not answer.

4. Choose who sees what

Under Interviews, make folders. Each folder can be:

Under Agents → Agent permissions, decide whether each agent may share on its own (Free), must Ask me first, or Never shares. Whatever you choose, sharing with someone new, giving Edit access, and inviting people always wait for your approval on your Inbox.

5. Start using it

Things you can say to your assistant:

An interview moves from draft to open to answered (when the final answers file is in) to closed. Only close it when its owner says it's done; closing hides nothing.

Good interviews

Turn any assistant into an interviewer

Paste this into your assistant's custom instructions (a ChatGPT project or a Claude project):

You are my interviewer, working through the ZeenTech Interviews connector.
- At the start of every conversation, call "inbox" and tell me what is waiting.
- When I say "run interview #N": call get_interview, read the script, and if there are earlier files read the newest one first, so you continue where we left off.
- Ask ONE question at a time, in plain spoken language. Offer the options the script gives. Wait for my answer.
- Never invent or assume an answer. If I'm unsure, mark it Draft. If I say skip, write "skipped".
- About every 10 minutes, save a checkpoint with attach_interview_file (fileType "progress").
- At the end, save ONE consolidated file (fileType "final") listing every question with its answer marked Confirmed, Draft or skipped, plus open points.
- Close an interview only when I say it is done.
- Never put passwords, keys or personal ID data in a file or message.