How a briefing document gives AI the context it needs every time

Executive overview

Every new AI chat thread starts blank — no knowledge of who you are, your role, your tone, or your goals. A briefing document solves this by packaging that context once, so you upload rather than retype it each time.

Keep 2–3 focused documents (one per use case) under 5 pages each. A smaller document leaves more context-window memory for the actual conversation.

The first draft becomes useful immediately, not after several rounds of correction.

What to put in a briefing document

• Your role, company, team structure, and KPIs
• How you want to work with AI (tone, voice, style preferences)
• Specific language rules (e.g. Australian English spelling, no filler phrases)
• Project-specific stakeholders, goals, and communications context
• A "canary" instruction to detect when the AI stops reading it (e.g. "never write 'I hope this email finds you well'")

How to use it in a thread

• Upload the document at the start of each new chat thread
• Ask the AI to refer to the briefing document explicitly in your prompt
• If the AI ignores it, type "refer to the briefing document" — it will redo the response
• Re-upload if a very long thread exhausts the context window

How many documents to maintain

• Most people need 2–3 briefing documents, not one giant file
• One for general work, one for specific projects, optionally one for personal use
• Fewer documents reduce the risk of confusing the AI across contexts
• Split any document that grows beyond 5 pages

How briefing documents relate to custom GPTs and projects

• A briefing document works across all AI tools, including those without paid plans
• Upload it as the knowledge source for a custom GPT, a Claude Project, or a Gemini Gem
• Useful for getting a second opinion: upload the same document to a different AI for comparison
• Works where platform-specific tools (custom GPTs) are unavailable or behave differently