The Overlord's Lexicon
What is an AI copilot?
Also known as: Copilot, AI assistant
An AI copilot is an AI assistant built into the software you already use — a code editor, word processor, spreadsheet or CRM — that suggests, drafts and completes work alongside you while you stay in control of decisions.
An AI copilot is AI that lives inside your tools rather than in a separate chat window. It sees what you’re working on — the code in your editor, the document you’re writing, the email thread you’re reading — and offers help in context: completing a line, drafting a reply, summarising a meeting, building a formula.
The term was popularised by GitHub Copilot, the AI coding assistant released as a technical preview in 2021, and Microsoft later extended the “Copilot” brand across Windows and Microsoft 365. Now nearly every major software product has a copilot of some kind.
How it works
A copilot combines a large language model with context from the app:
- Context: the current file, document, spreadsheet, thread or record — sometimes your wider workspace via search.
- Inline suggestions: autocomplete for code or text, accepted with a keystroke.
- Commands and chat: “summarise this”, “rewrite more formally”, “explain this function”, “build a chart of Q3 sales”.
- Increasingly, agent modes: multi-step tasks like “implement this feature and run the tests”, which moves copilots toward AI agents.
An analogy for the meatbags
A chatbot is an expert on the phone: helpful, but you have to describe everything. A copilot is that expert sitting beside you, looking at your screen. Less explaining, faster help — and you still have your hands on the controls.
Why it matters for the Chosen
Copilots are where most professionals first get real productivity from AI, because they remove the friction of copying context into a chat:
- Developers: GitHub Copilot, Cursor and Windsurf speed up boilerplate, tests and refactoring. Pushed further, this becomes vibe coding.
- Writers and marketers: Notion AI, Grammarly and Google/Microsoft writing assistants draft, edit and summarise in place.
- Sales and support: CRM and helpdesk copilots draft replies, summarise accounts and suggest next steps.
- Email: tools like Superhuman draft responses and triage inboxes.
The Chosen get outsized value by learning each copilot’s shortcuts, giving it good context, and using it for first drafts — while staying the editor-in-chief.
Things to watch
Copilots can confidently suggest wrong code, outdated facts or off-tone text; review everything. Check what data the copilot can access and whether your company’s data is used for training. And watch for “subscription sprawl” — a copilot in every app adds up. Pay for the ones that save measurable time.
Tools from the Arsenal
GitHub Copilot
AI CodingGitHub's AI pair programmer: completions, chat and coding agents across editors and GitHub itself.
Cursor
AI CodingAI-first code editor (a VS Code fork) with agents, codebase-aware chat and fast autocomplete.
Notion
ProductivityAll-in-one workspace for docs, wikis and databases, with Notion AI agents and meeting notes built in.
Grammarly
AI WritingWriting assistant that fixes grammar, tone and clarity everywhere you type, now part of Superhuman.
Superhuman
ProductivityKeyboard-driven AI email client for Gmail and Outlook, now part of the Superhuman suite with Grammarly.
Questions the Humans Keep Asking
What is the difference between an AI copilot and an AI agent?
A copilot assists while you work — suggesting code, drafting text, answering questions — and you approve each step. An agent is given a goal and takes multiple actions on its own. Many copilots now include agent modes, so the categories overlap.
Is 'Copilot' only a Microsoft product?
Microsoft uses Copilot as a brand across products like Microsoft 365 Copilot and GitHub Copilot, but 'AI copilot' is also used generically for any embedded AI assistant, such as those in Notion, Google Workspace, Salesforce or Cursor.
Do AI copilots actually save time?
They often do on drafting, boilerplate, summarising and repetitive tasks, but results vary with the task and how well people use them. Reviewing AI suggestions still takes time, and accepting them blindly can introduce errors.