Ask Copilot today how many vacation days you have left and you will most likely get a well-written paragraph with three links. One to the HR portal, one to the leave policy and one to a form that asks, once again, for your name and department. Copilot understood the question. You finish the job yourself, in another tab.
This morning Microsoft announced the general availability of Copilot UX components. The idea is simple: when the conversation calls for it, a screen shows up instead of a paragraph. Your remaining days, the proposed dates and a button to submit the request, all inside the chat.
And the most interesting part is not the screen. It is what it is built with.
From wall of text to button
A UX component is a small web application that Copilot shows when the question needs it. Microsoft gives two examples that need no technical explanation:
- Someone asks how the quarter is going and a sales dashboard opens, with filters to explore the numbers.
- Someone wants to request time off and the balance, the dates and a submit button appear.
The component can show up in two ways: inline, inside the conversation, for quick actions, or full screen, when the task needs room, such as a dashboard, a multi-step form or a map. It also adapts to Copilot's light or dark theme.
The division of labor is clear. Copilot interprets what the person wants and calls the component. The component shows the data and options your code defines. And before any important change, the person reviews and confirms.
AI understands the intent. The screen does the work. The person decides.
The same web part, on a new stage
Here is the part that interests me as someone who builds SharePoint portals. These components are made with the SharePoint Framework (SPFx), the same tool that has been used for years to build custom web parts for corporate intranets. Microsoft estimates that SPFx solutions already serve tens of millions of people every day, across tens of thousands of organizations.
That changes the conversation for many companies. The KPI dashboard on the intranet home page, the request form or the people directory do not need to be rebuilt from scratch to reach Copilot. The same package can work as:
- a component inside Copilot,
- a web part on a SharePoint page,
- a tab in Microsoft Teams,
- an app in the Outlook side menu.
One codebase, four entry points. People choose how they get there.
No new server to look after
Under the hood, the components use the MCP Apps model, an open standard for showing interfaces inside AI assistants. If you are not sure what MCP is, we explained it in this article.
Any company could build one of these apps on its own and host it in Azure. The difference is who carries the operations:
| Self-hosted MCP app | Copilot UX component | |
|---|---|---|
| Where it lives | Your own server, in Azure or another cloud | Inside the company's own Microsoft 365 |
| Sign-in | You build and maintain it | Automatic, with the user's own permissions |
| IT approval | A new external service to review | The same process already used for SPFx |
| Maintenance | Patching, monitoring and uptime are on you | No extra infrastructure to run |
| Reaching many customers | One deployment per customer, or your own platform | One package installed in each tenant |
For an IT team, the approval row is worth gold. No new vendor to assess and no extra server in the inventory. The solution is installed in the SharePoint app catalog like any web part, and it respects the permissions each person already has.
Three paths to the same list
In just a few weeks Microsoft showed three different ways to build on top of SharePoint data with Copilot's help. First HTML pages, then Cowork apps, and now this. It helps to put them side by side:
| HTML page | Cowork app | Copilot UX component | |
|---|---|---|---|
| Who builds it | Any author, or Copilot | Any user, by chatting | A developer |
| Where it shows up | On a SharePoint page | In its own portal, outside SharePoint | In the Copilot chat, plus SharePoint, Teams and Outlook |
| Data | Read only | Read and write through connectors | Read and write through any API |
| Knows who is using it | No | Yes | Yes |
| License to use it | None | Power Apps Premium or credits | Copilot license |
| Maturity | Preview | Frontier preview | Available as of today |
None of them replaces the others. The HTML page is for showing something fast. The Cowork app is for an afternoon prototype. The UX component is for the process the company will run for years, with versioned code and someone accountable for it.
The fine print
- It requires a Copilot license. During the preview anyone could try them. Now, users without the Copilot add-on license do not see the component, only a message explaining why. For companies with few licenses, this decides who the experience reaches.
- No usage charges. The component itself does not consume credits. If it calls a service billed by consumption, that service bills separately.
- Not on mobile yet. It works in Copilot in the browser and on desktop, and also inside Teams. Android and iOS are on the to-do list.
- It lives inside an agent. For now the component ships with a specific agent, and the person has to be using that agent. Microsoft plans to make it appear directly in the regular Copilot chat, and in Cowork too, but with no date.
- It needs a developer. This is not built by chatting. It is written in React, TypeScript or JavaScript, tested and deployed. The good news is that the people who already maintain the intranet web parts are the same people who can build it.
Where to start
Microsoft's recommendation makes sense and does not require a big project: pick a single task that people already ask Copilot about but finish in another portal. A time-off request, an expense approval, a quick KPI check. That is your first candidate.
And before writing any code, one useful question: which web parts on your intranet get the most visits? If people already open the portal every day to do the same thing, that is the component that will save the most time once it shows up in the chat.
The point
For years the intranet was a place you had to go to. With Copilot UX components, it starts coming to the conversation. And for companies that invested in SharePoint, the good news is that the investment did not go stale. It just changed stages.
If your company has web parts worth bringing into Copilot, or you want to know whether this path makes sense with the licenses you already pay for, reach out and let's talk.
Sources
- Copilot UX components are generally available, Microsoft 365 Developer Blog
- SharePoint Framework (SPFx) roadmap update, September 2026, Microsoft 365 Developer Blog
- Going beyond text in Microsoft 365 Copilot, Microsoft 365 Developer Blog
- Overview of Copilot UX components, Microsoft Learn
