What Copilot in PowerPoint can actually do in 2026

Most of what has been written about Copilot in PowerPoint falls into one of two camps. One says it will make every deck for you. The other says it produces generic slides nobody would present. Neither is a useful basis for a decision, and neither describes what the product does today.

This guide is the version we give clients: what Copilot in PowerPoint does, what it depends on, and where a person still has to step in. Every capability below is drawn from Microsoft’s own documentation, linked at the end, as of September 2026. The product changes often, so check the dates.

1. It drafts a presentation from a prompt

The headline feature is still the simplest one to describe. You open PowerPoint, open the Copilot pane, and describe the presentation you want. Copilot may ask clarifying questions — who the audience is, what presentation and image style you want — and then proposes an outline. You can edit that outline before any slides are made, which is the most underused step in the whole process. A good outline is most of a good deck; a bad one produces twelve well-formatted slides of the wrong argument.

Once you accept the outline, Copilot generates the slides, and you can keep chatting with it to add, remove or rework them.

Two constraints are worth knowing before you plan around this:

  • Microsoft states plainly that the output “is AI-generated content and should be human-reviewed and edited accordingly.” That is not boilerplate. It is the correct operating assumption.
  • Only one output language is supported at a time.

2. It works from your source documents

Copilot can build from existing material rather than a blank prompt. Microsoft documents support for referencing files — a Word document under 24 MB, for example — and images from those documents are used where available. In the newer Agent Mode experience, Microsoft lists file referencing as “coming soon,” so the exact path depends on which experience your organization has.

In practice this is where most of the value sits. Very few corporate decks start from nothing. They start from a report, a plan, a proposal or last quarter’s version. Turning that material into a structured first draft is exactly the kind of work Copilot is suited to.

3. It follows your template — to the degree your template teaches it

This is the capability that matters most to a brand or creative team, and the one least understood.

Microsoft recommends starting from your organization’s standard template so the theme and design carry through. But Copilot does not simply copy the look of a template. It learns from it. Microsoft’s guidance is explicit about what it learns from:

  • Sample slides are the primary teaching material. Microsoft recommends representative examples across the slide types you actually use — title, agenda, section divider, content, data, timelines, process, quotes, summary and so on — at light, medium and heavy content densities.
  • Placeholders tell Copilot where content goes. It uses the placeholder’s type, position and size. It does not read the words inside a placeholder as instructions.
  • Theme definitions — theme colors and heading and body fonts — carry your brand into everything Copilot generates.

And one sentence from Microsoft’s documentation deserves to be on every brand team’s wall: “If your template does not contain enough sample slides, Copilot falls back to the Slide Master — which may result in less optimal output.”

In other words, Copilot’s output is capped by the quality of the template it is given. We come back to this idea often in this series, because it changes who is responsible for the quality of your company’s decks.

4. It uses your Brand Kit

A Brand Kit is a central collection of your brand assets inside the Microsoft Copilot app: logos, color palettes, fonts, approved imagery, templates, and even brand voice and photography guidance. When someone selects a brand kit in the Copilot pane before prompting, Copilot uses that kit’s approved templates, colors, fonts and assets.

Official Brand Kits — the ones shared with the whole organization — can only be created and edited by people your IT team designates as brand managers. That is deliberate. It puts control of what Copilot treats as “the brand” in the hands of a named owner.

5. It can be locked to your approved layouts

In September 2026, Microsoft added two template settings to Brand Kit:

  • Strict brand adherence. When it is on, Copilot uses only your approved template and slide master layouts. It cannot add or remove placeholders or invent new layouts. When it is off, Copilot may reuse some of your layouts or “get more creative,” while still applying your colors, fonts and assets.
  • Note instructions (notes steering). Brand managers can write plain-language instructions in a template’s speaker notes — “Always include this cover slide,” “Don’t modify this slide,” “Keep the text minimal” — and Copilot follows them.

Only the Brand Kit owner or a brand manager can change these settings, and employees creating presentations from the template cannot override them.

This is a significant change. Before it, a template was a suggestion. Now it can be a rule. That makes the template far more powerful, and it makes a weak template far more expensive, because every employee is now held to it.

6. It can follow written rules: Brand Skills

Also in September 2026, Microsoft began rolling out Brand Skills: brand managers can upload instructions as Markdown files to Brand Kit and publish them to the organization, so Copilot in PowerPoint follows them when creating presentations. Think of it as your presentation standards written in a form the model reads. Microsoft is clear that this is not switched on automatically — someone has to write, test and publish the skill — and it recommends governance review and periodic audits of published skills.

What it cannot do for you

Knowing the limits is what separates a useful rollout from a disappointing one.

  • It cannot decide what you are trying to say. Copilot will structure the material you give it. It will not tell you that the real recommendation is buried on slide nine, or that the audience needs the conclusion first.
  • It cannot fix a template that teaches the wrong thing. Hardcoded colors, text boxes instead of placeholders, no example slides — Copilot inherits all of it.
  • It cannot guarantee brand conformance. Strict mode constrains layouts; it does not make a generative model deterministic. Review still matters, especially for customer-facing and executive work.
  • It does not remove the need for judgment on the decks that matter. A board presentation, a keynote, a pitch — the stakes are not in the formatting.

What this means for your team

The practical shift is this: for routine internal decks, the person doing the work is increasingly Copilot, and the thing directing it is your template, your Brand Kit and your written rules. The quality of those three assets now sets the floor and the ceiling for thousands of presentations you will never see.

That is why we think the most important Copilot decision a brand team makes is not which prompts to teach people. It is whether the template is built for a model to learn from, rather than only for a person to use.

Key takeaways

  • Copilot drafts decks from prompts and from source documents, and you can shape the outline before any slides exist.
  • It learns from your template’s sample slides, placeholders and theme — and falls back to the Slide Master when examples are missing.
  • Brand Kit, strict brand adherence, notes steering and Brand Skills give brand managers real, central control over what Copilot produces.
  • Microsoft’s own position is that output should be reviewed by a person. Plan for that.

Want to know how your current template would teach Copilot? Have your template scored — we review it against Microsoft’s published guidance and tell you where it stands.

Sources

Blog

Other popular articles

24×7 Design Services

Get a quote now