Business Central 29 Ships in October Without a Release Plan: How to Scope the Regression Pass

What changed this month, and what did not

Microsoft's partner announcement of 4 September is unambiguous. Roadmap content for Dynamics 365, Power Platform and Dataverse is moving to the AI at Work roadmap, and, in the announcement's own words, "there won't be a September 2026 Release Wave 2 announcement. Instead, Microsoft will archive Release Plans on Microsoft Learn and retire Release Planner by November 15, 2026."

The Business Central documentation says the same from the product side: from September 2026 Microsoft stops publishing Release Plans, new capabilities go to the AI at Work roadmap, and features are no longer linked to release plans at all.

That is a change to how updates are announced, not to how they ship — and the distinction is the whole story for anyone running a test cycle.

The update still arrives on the old schedule

Coverage since August has blurred the two, leaving the impression that Business Central itself is going continuous. It is not. Microsoft put it plainly when it announced the roadmap move: the transition "changes how Microsoft communicates upcoming innovation. It doesn't change how our products are built, released, or deployed."

The mechanics are unchanged:

The deadline your customers face in October is the one they faced last October. What has gone is the artefact you planned against. We covered that shift when the release wave model was retired; 29.0 is where the consequences land.

Where the feature list lives now

Two sources replace it, and they are not equivalent. The "What's new or changed" article for update 29.0 on Microsoft Learn is the closer substitute and the better one for testing: dozens of entries across development, finance, supply chain, reporting and Copilot, tied to a version number rather than a wave. The AI at Work roadmap is the forward-looking feed — items moving through In Development, Rolling Out and Launched, filterable by product and exportable to CSV.

A roadmap built around always-on publishing tells you what is coming. It does not give you a bounded, dated list of what is in the version landing on a given tenant next month. Rebuilding that list is partner work now. Under the wave model the job had a natural twice-yearly rhythm; under continuous publishing it has none, which in most partner businesses means it quietly becomes nobody's.

Your preview sandbox has a deletion date

The 29.0 preview has been available in sandbox environments since the first week of September. Microsoft's documentation also states that it automatically deletes preview sandboxes 30 days after 29.0 goes generally available — early November 2026.

This catches teams out every wave. A partner spins up a preview sandbox, runs a partial regression pass, logs a few defects, and plans to finish once the customer's own upgrade is scheduled in January. By then the environment is gone, along with the test data and configuration that made the results meaningful. Evidence of a pass that lives inside the sandbox expires on the same schedule.

Scoping a regression pass without a release plan

The release plan was always a weak basis for scope: it described what Microsoft built, not what your customer uses. Losing it is an argument for scoping the way it should have been all along.

Anchor the pass to the version, not the announcement. The trigger is "this tenant is scheduled for 29.0 on this date", taken from the admin centre — a fact that still exists and is still dated, which is more than the roadmap gives you.

Build the scope from the customer, then check it against the version. Start with the processes that stop the business if they break — order to cash, the month-end close, the integrations, the reports finance actually opens — and run those every wave regardless of what Microsoft shipped. Then read the deprecation list and the 29.0 article for anything touching them. Reading the release plan first tended to produce a test pass shaped like Microsoft's backlog.

Keep a reusable regression suite per customer. Rebuilt from scratch each October, the missing release plan is a real problem. Maintained and topped up with version-specific cases, it is an inconvenience. Our Business Central UAT guide covers how to structure one, and what separates regression from acceptance testing is worth settling before October.