About Platform Resources Articles Contact
IMA AI  ·  Resources

Using AI to Build
Project Documentation

Every time a staff member explains a product, a client brief, or a project scope to someone new — that is wasted time. This guide teaches you how to use AI to turn your work discussions into structured documentation that the whole team can reuse.
Version v1.0
Date May 2026
Scope All IMA AI Staff
Tools Claude · ChatGPT · Gemini · Any AI
Context

Why This Exists

The same explanation happens again and again because nothing was written down properly.
The core idea: When you discuss a topic with an AI, you are not just getting an answer. You are building a record. If you discuss it properly, at the end you can ask the AI to write everything confirmed into a clean markdown file. That file becomes a reusable asset for the whole team.
Discussion with AI
       ↓
Confirm facts, decisions, details
       ↓
Ask AI to write the MD file
       ↓
Submit MD to project lead
       ↓
Gets combined into master project file
Part 1

Before You Start

RuleOne topic per chat
Never mix two different topics in one chat session. Each chat should cover one focused scope. If you realise mid-chat that you are drifting into a second topic — stop. Finish the first topic, get the MD file, then start a new chat for the second topic.
Wrong ❌Right ✅
All products + website + video script in one chatOne product's details in one chat
Homepage + about page + services all at onceHomepage copy only
Script structure + product spec + pricing togetherVideo script structure only
RuleGive context first
Never open a chat with just a question. Give the AI the context it needs to help you properly. The more context you give at the start, the better the output at the end.
Wrong ❌

"Write me a product description."

Right ✅

"I am working on [product name], a [what it does] for [who it is for]. The brand tone is [describe tone]. I need to document this product so our team can use it for the website, video scripts, and other content. Let me share what I know and you help me confirm and structure it."

RuleKnow what you want at the end
Before you start the chat, know what MD file you are trying to produce. Examples: Product reference MD · Video script template MD · Campaign brief MD · Service scope MD
Part 2

During the Chat

RuleConfirm, don't assume
As you discuss, make sure you confirm facts explicitly. If something is agreed, say it clearly so the AI records it.
"So to confirm — the product targets people with [condition]. The three key features are [A], [B], and [C]. The tone is [describe]. Is that correct?"
RuleFlag what is pending
If something is not yet decided, say it clearly. Pending items must be captured, not ignored. They are the things that fall through the cracks.
"We don't have the price confirmed yet. Flag this as pending in the MD file."
"The store URL is not ready. Mark as pending — update when confirmed."
RuleStay in scope
If the AI starts going off-topic or getting too detailed in the wrong area, redirect it.
"Let's stay focused on the product spec for now. We'll do the video script in a separate chat."
Part 3

Ending the Chat

When the discussion is complete and everything is confirmed, ask the AI to write the MD file.
TemplateHow to ask for the MD file
Be specific about what you want.
"Based on everything we confirmed in this chat, write me a clean markdown file for [topic]. Include all confirmed details. Add a Pending section at the bottom for anything not yet confirmed. Use clear headings so it is easy to read and update later."
ChecklistWhat a good MD file includes
1Title and metadataWhat it is, date, version, status
2Confirmed contentEverything discussed and agreed, organised by section
3Pending itemsClearly listed at the bottom, nothing buried
RuleReview before submitting
Before you submit the MD to your project lead, read through it once. If something is wrong, correct it in the same chat — then ask for the updated version.
"In the MD file you wrote, [detail] is missing. Please add it."
Part 4

Submitting the MD

RuleWho you submit to
Submit your MD file to whoever is managing the master project file for that product or project. When in doubt, submit to the project lead or CEO.
TemplateHow to submit
Send the MD file directly. Include a one-line note about what it covers and what is still pending.
"Attaching [topic] MD. All [details] confirmed. Pending: [what is missing]."
What happens next: The project lead reviews your MD and merges it into the master project file. Your work becomes part of the permanent reference that the whole team uses — for briefings, content creation, onboarding new staff, and any task that needs context about that product or project.
Part 5

Real Examples

Three common scenarios showing what to confirm in the chat and what MD file you produce.
Example 1
Product Reference MD
Situation
You need to document a product so the team can use it for the website, video scripts, and social content without having to ask you every time.
Chat scope
One product only
Confirm in chat
Product name, series, category · Who it is for (target audience, conditions, use case) · Key features or ingredients and what each does · Brand tone and copy rules · Proof points (certifications, research, awards) · What is pending (pricing, store URL, final copy)
MD file produced
[brand]-[product-name]-product.md
Used for
Website product page, video script, social captions, briefing new staff
Example 2
Video Script Template MD
Situation
You want a standard video script structure that any staff member can follow for a brand's product videos.
Chat scope
Script structure only — not the actual script content
Confirm in chat
Video length and format · Opening hook approach · Section order (problem → solution → proof → CTA) · Tone rules (what to say, what never to say) · CTA format and placement
MD file produced
[brand]-video-script-template.md
Used for
Every video produced for that brand — staff follows the template, no briefing needed every time
Example 3
Campaign Brief MD
Situation
You need to brief the team on an upcoming campaign.
Chat scope
This specific campaign only
Confirm in chat
Campaign dates and hours · Products or services featured · Pricing and promotions · Platform or channel · Staff assigned and their roles · What is pending (final product list, confirmed promo details)
MD file produced
[brand]-[platform]-[campaign]-brief.md
Used for
Briefing all staff involved — one file, no repeated briefings
Part 6

File Naming

Name your MD files using the same standard as all IMA AI files — lowercase, hyphens only, descriptive.
Pattern: [brand]-[topic]-[type].md
FileWhat it covers
[brand]-[product-name]-product.mdProduct reference for one product
[brand]-video-script-template.mdVideo script standard structure for a brand
[brand]-[platform]-[campaign]-brief.mdCampaign brief for one campaign
imaai-[service-name]-scope.mdIMA AI service scope reference
Part 7

Rules Summary

RuleWhy
One topic per chatKeeps the MD file clean and focused
Give context firstBetter context = better output
Confirm explicitlyWhat is not confirmed is not reliable
Flag pending itemsPending items disappear if not captured
Review before submittingYou are responsible for accuracy
Submit with a one-line noteProject lead needs to know what's in it and what's missing
No client names in examplesKeep internal docs free of client references
Any AI tool worksThe workflow is the same regardless of which AI you use
Pending

Outstanding Items

Add AI tool-specific notes — how to start a focused chat in Claude vs ChatGPT vs Gemini
Add a completed product MD as a reference example for staff
Translate key sections to Bahasa Malaysia if needed for non-English-fluent staff