AI Projects

AI Projects keep persistent instructions, files, decisions, and context around a workstream so teams do not restart every chat from zero.

• A persistent workspace for a product, campaign, client, initiative, or operating process.
• Keeps instructions, reference files, decisions, and recent context together.
• Reduces repeated setup and makes handoffs easier.

• Do not use for quick one-off questions.
• Do not use as a dumping ground for every file.
• Do use when the same context will matter across multiple sessions or people.

• Create one project for one workstream.
• Add only the files and instructions that materially change output quality.
• Capture decisions and constraints in a short project note.
• Review the project monthly and remove stale context.

• Keep project boundaries clear.
• Prefer current source files over copied stale excerpts.
• Name the project after the outcome, not the tool.
• Related: Use MCP to connect AI assistants to real work context