What it does

"Copilot" in the Power Platform is a set of maker-facing generative AI features built into Power Apps, Power Automate and Power Pages that let you describe what you want in natural language and get an app, flow, page, form or table back. It's a build-time assistant for makers, not the same thing as Copilot Studio, which is the separate product for building the agents your users talk to.

Key facts

• Copilot shows up per product: build canvas apps and app skills in Power Apps, draft cloud flows in Power Automate, and generate sites, pages, forms, text and themes in Power Pages.
• These features share one Microsoft account (the Learn hub at the Source URL) but each has its own docs, availability and rollout state, so don't assume parity between products.
• Copilot Studio is the distinct low-code product for building agents (formerly Power Virtual Agents plus the copilot authoring canvas). The maker Copilots help you build; Copilot Studio builds the thing users chat with.
• AI Builder is the other piece of the stack: prebuilt and custom AI models (document processing, prediction, prompts) you call from apps and flows. It underpins some Copilot features but is licensed and managed separately.
• Availability varies by geography, language and tenant setting. Some features are GA, others preview; the international availability report (aka.ms/bapcopilot-intl-report-external) is the source of truth.
• Admins control all of this centrally through the Copilot hub and Power Platform admin centre governance settings, including turning generative AI features off per environment or tenant.
• Data handling follows the Power Platform Copilot data security and privacy FAQ; some features have an optional data-sharing toggle that's separate from the on/off switch.

When to use / skip

Use the maker Copilots to get a first draft fast: a canvas app from a description, a flow from a sentence, a Power Pages site scaffold. They're good for accelerating the blank-page stage and for makers who don't want to hand-build every screen. Skip the assumption that Copilot output is production-ready. Treat it as a starting point you refine, not a finished solution. If the requirement is a conversational agent your end users interact with, that's a Copilot Studio job, not a maker Copilot one, so start there instead. Don't reach for this category when the real need is a custom AI model on documents or predictions, which is AI Builder.

Configuration decisions

• Decide per environment whether generative AI features are on at all, using the Copilot hub and admin centre governance controls, before makers start relying on them.
• Work out the optional data-sharing setting separately from the on/off decision, and get sign-off from whoever owns data governance.
• Check geographic and language availability for each specific feature you plan to use rather than assuming your tenant has all of it.
• Decide where agents live: maker Copilots build apps and flows, but anything conversational belongs in Copilot Studio with its own licensing and capacity.
• Confirm AI Builder licensing and capacity if any Copilot or app feature you want depends on AI Builder models.

Gotchas

• People conflate the maker Copilots with Copilot Studio and with Microsoft 365 Copilot. Three different things. Pin down which one a stakeholder means before you quote effort.
• Feature parity across Power Apps, Power Automate and Power Pages is uneven, and preview features move fast, so a demo you saw last quarter may not match what's in the tenant today.
• Turning Copilot off is a governance decision with real blast radius; makers who've built habits around it will notice when it disappears.
• The naming keeps shifting (Power Virtual Agents became part of Copilot Studio, "copilots" became "agents"). Older docs and blog posts use terms that no longer match the UI.
• Availability by region and language is genuinely restrictive for some features; check before promising a client in a non-English or non-US environment.

Consultant notes

• This doc is the category's front door. Send people here first, then to the per-product docs (Power Apps, Power Automate, Power Pages) and to the separate Copilot Studio category for agents.
• When scoping, always split "help the maker build faster" (this category) from "give users an agent to talk to" (Copilot Studio). They price, license and govern differently.
• The single most useful link to keep handy is the international availability report; it settles most "can we actually use this here" arguments.
• Watch this one: Microsoft rebrands and reshuffles these features often. Date-stamp anything you tell a client and re-check the Learn hub before a big commitment.

---

Re-check feature GA/preview status and the availability report before this goes in front of a client. Microsoft moves the goalposts on this category constantly.