SharePoint Copilot Apps and SPFx 1.24: what is coming in 2026

In July 2026 Microsoft started the public preview of SharePoint Copilot Apps. General availability is announced for September 2026 together with SPFx 1.24, including the SPFx CLI and navigation customizers. For organisations running SharePoint as their intranet, this is the biggest change to the development model in years.

The decisive phrase in Microsoft's roadmap is that solutions should be “Copilot-ready by default”, surfacing data and actions directly in Microsoft 365 Copilot. In plain terms: what you build as a web part today can appear on the Copilot canvas in Teams, SharePoint and Viva tomorrow. Existing SPFx knowledge is not devalued – it gains a second stage.

What are SharePoint Copilot Apps?

Until now the split was clear: a web part lives on a SharePoint page, and Copilot is a chat window working from general knowledge and indexed documents. Between the two sat a gap – the agent did not know your processes, and the web part could not answer questions.

SharePoint Copilot Apps close exactly that gap. You still build with the SharePoint Framework, but you additionally declare which data and which actions your solution offers to Copilot. The agent can then not only answer from your content but also trigger something: start an approval, create a record, kick off a workflow.

The practical difference is the entry point. Your people no longer need to know which intranet subpage holds the right tool. They ask where they already ask – and the answer comes from your own data, with a source reference and within existing permissions.

The roadmap: from SPFx 1.23 to 1.24

Microsoft has moved SPFx onto a more predictable, quarterly-oriented rhythm. The relevant milestones of 2026:

What this means for existing web parts

The reassuring part first: there is no announced deprecation of classic web parts. You do not have to throw anything away. Two items still belong on the agenda.

First, the React 18 migration. Custom web parts can adopt React 18 today; broad enablement is tied to the completion of the out-of-the-box templates and planned for the SPFx 1.24 GA. If you run a larger web part estate, schedule this migration rather than inheriting it.

Second, how you scope new web parts. If you are building something anyway, it pays to separate data access and actions cleanly from presentation. That separation later decides whether a solution can be surfaced as a Copilot app with modest effort – or has to be rewritten. The extra cost today is small; the saving later is substantial.

SharePoint Server 2019: the other date this summer

In parallel with the Copilot movement, support for SharePoint Server 2019 ended on 14 July 2026. For organisations still running on-premises, that is the harder deadline: there are no more security updates.

The difficult part of a move is rarely the data migration – there are tools for that. What is critical are the customisations that grew over the years: JSLink, sandbox and farm solutions, InfoPath forms, SharePoint Designer workflows. None of these run in SharePoint Online. If you only copy files, you discover after the move that exactly the processes your departments depend on every day are gone.

That is why a SharePoint migration sensibly begins with an inventory analysis: what is genuinely in use, what has been dead for years, and what has to be rebuilt as an SPFx web part? In our experience the list is far shorter than feared.

What organisations should do now

Both movements together produce a fairly clear plan for the second half of 2026:

1. Take inventory. Which web parts and customisations run in the tenant, who uses them, which SPFx version are they on? Without that list, any planning is guesswork.
2. Schedule the React 18 migration. Set a date for existing custom web parts instead of waiting to be forced.
3. Pick one Copilot use case. Not ten. One that answers a frequently asked question and whose data source is properly maintained.
4. Review permissions. An agent mercilessly exposes libraries that are shared too openly. That is uncomfortable – but better before the rollout than after.
5. Prepare documentation. Purpose description, risk classification and labelling are obligations under the EU AI Act anyway.
6. Resolve on-premises. If you are still on SharePoint Server 2019, prioritise the migration – running a document archive without security updates is not a tenable state.

And one caveat few people say out loud: not every process needs custom code. If your use case can be solved with a SharePoint list and Power Automate, that is faster, cheaper and easier to maintain. A custom web part only pays off once the built-in tools demonstrably cannot carry the case.

Frequently asked questions

Are classic SPFx web parts obsolete now?

No. Microsoft has announced no deprecation of classic web parts; the roadmap updates treat Copilot Apps as an additional surface rather than a replacement. Existing web parts keep running. The one thing to plan for is the React 18 migration, whose broad enablement is scheduled for the SPFx 1.24 GA.

Should we wait for GA in September 2026?

For the production rollout of a Copilot app there is a good case for waiting until general availability. For preparation and a pilot the opposite holds: inventory, permission review and data preparation take weeks anyway and are independent of the version. Doing that work now means you are ready at GA rather than starting then.

What does a SharePoint web part cost?

At Lyron a single, clearly scoped web part starts at €1,900 as a fixed price. An intranet suite of three to five web parts including a design system starts at €5,900, and a Copilot app at €9,900. We give the binding price after the intro call, once the scope is clear.

Does the Copilot agent see all documents?

It should see exactly what the person asking is allowed to see – no more. That is precisely why reviewing permissions before rollout matters: an agent surfaces overly broad sharing that nobody noticed before, simply because nobody went looking for it.

We are still on SharePoint Server 2019 – how urgent is this?

Urgent. Support ended on 14 July 2026 and there are no more security updates. For a system that typically holds contracts, HR records and project documentation, that is a real risk rather than a formal footnote.