Turn “ask the owner” into a useful handbook.
You answer the same question every month. Luminair turns how you actually did the task into a guide a new hire can follow alone.
Here is a recording of me doing this task, three support chats where we handled it, and our checklist. Write a step by step guide a new hire can follow alone, including the unusual cases, and add a short practice exercise at the end.
Claudewrites the steps from your real examples- Geminitests them against the unusual cases
Codexturns it into a searchable handbook page
Handbook: refund requests
- Drafted from 3 resolved Intercom tickets + 1 walkthrough
- Exception found: partial refund needs manager sign-off
- Tested by a second team member, 2 steps clarified
The same question comes up every few weeks, and the answer lives in your head. A new hire asks it again, you explain it again, and nothing gets written down because writing it down felt like a separate project.
- The same question gets answered the same way, out loud, every month
- Institutional knowledge lives in one person’s head
- Onboarding takes as long as it did for the last three hires
Try it yourself in Luminair
Six steps. No setup beyond the AI plan you already have.
- 1
Install Luminair and sign in
Download Luminair and open it. Go to Settings → Models → Add account and sign in with an AI plan you already pay for, like Claude, ChatGPT or Gemini.
- 2
Make a folder for this job
In the sidebar, choose New folder… and name it “Handbook”. Then choose New session in this folder. Everything for this job stays together.
- 3
Bring the information in
You need: a recorded walkthrough, resolved support chats and your checklist. Connect Fathom, Intercom and Asana in Settings → Connectors, or export the files and drag them into the chat.
- 4
Ask for the result
Paste this into the session, or say it in your own words:
Here is a recording of me doing this task, three support chats where we handled it, and our checklist. Write a step by step guide a new hire can follow alone, including the unusual cases, and add a short practice exercise at the end.
- 5
Get a second opinion
Pick a different model in the model picker and ask it to check the first answer:
Try to follow this guide for the hardest of these chats. Where would a new hire get stuck?
- 6
Make it repeat
Open Flows → New and add the steps: Read newly resolved support chats → Suggest changes to the guide → Flag steps that contradict each other. Choose when it runs, like every Monday at 09:00. Nothing is sent or changed until you approve it.
How it plays out
Start from how it was actually done
Make a folder called Handbook. Drop in a recording or notes of you doing the task, plus three past examples, including one that went unusually.
Get the guide, then test the edges
Ask for numbered steps a new hire can follow alone. Then ask a second model to walk through the hardest example with it and point out where it breaks.
Have someone use it for real
Ask a colleague to do the task from the guide and fix what was unclear. Save the final version in Journal so every session can answer from it.
You stay in charge
You separate real policy from a one-time workaround — that judgment stays with the owner, not the draft.
Next quarter’s question gets answered by the handbook, not by interrupting you.
Sign in with the AI plan you already have and start with this job.