Business Central 2026 release wave 2: what finance leaders should turn on, and what can wait

Business Central 2026 release wave 2, version 29, is rolling out now. Most of the coverage is about AI agents and the new Copilot. Those matter, but they aren't the changes most likely to help your finance team this quarter.

The most useful items for a CFO or controller are quieter: new approvals, an automatic audit trail on financial reports, and better tools for checking what your reports actually contain. Here's what's in the release for finance, what to switch on, what to pilot, and what to leave alone for now.

First, the timing

Update 29.0 started the wave in the fall, and update 29.1 followed in November. If you're on Business Central online, your administrator can schedule the 29.0 update on any date within Microsoft's update window, which runs to February 2027. Don't let it slide past month-end or year-end close. Pick a quiet week, test in a sandbox first, and let it run.

One change in how Microsoft communicates: it stopped publishing release plans in September 2026. New capabilities now appear on Microsoft's AI at Work roadmap, and the "What's new" pages for each update are the most reliable list of what actually shipped. If your partner is quoting features from an old release plan, ask which update they're in.

Turn these on: controls your auditors will like

Approvals for fixed asset journals (update 29.1). Fixed asset journals can now require approval before entries post, so acquisitions, write-downs and disposals get the same sign-off as other journals. If your auditors have ever asked who approved a fixed asset adjustment, set this up. It takes an afternoon, not a project.

Batch totals for payment journals (update 29.1). A small change that helps whoever releases payment runs: you can see the batch total before you post. It's the kind of check people currently do in a spreadsheet.

Financial report changes are always logged (update 29.0). This one needs no setup at all. Business Central now records every change to financial report definitions: the report, its rows and its columns. You can see what changed, who changed it and when in the Change Log. Separately, the Financial Report Audit Logs page shows who ran which report, when, and whether they exported it to Excel or PDF. (Microsoft's documentation.)

This closes a real gap. It's common to find financial reports where a row definition was quietly edited months ago and nobody can say why the numbers moved. Now there's a record.

Trace G/L account usage in financial reports (update 29.0). You can check which reports a given G/L account feeds into. Use it before you add, merge or retire accounts. It's also a fast way to find accounts that appear in no report at all, a common sign of a chart of accounts that has grown without a plan.

Run several financial reports into one PDF (update 29.0). Useful for the monthly board or lender pack. It's minor, but it removes a manual step at the busiest time of the month.

Pilot these: the Payables Agent and agent controls

The Payables Agent reads vendor invoices from email and drafts purchase invoices for a person to review. Version 29 adds two things finance teams asked for:

  • Better purchase order matching (29.0). The agent is better at matching invoices to the right purchase order lines.
  • Vendor credit notes (29.1). The agent can now process credit notes, not just invoices.

Around it, Microsoft has made agents easier to supervise. There's a dedicated task pane for everything every agent is working on, you can review agent-generated content directly on the page, lists show notifications when agent actions need review, and agent permissions are simpler to manage.

Our view: these are the right improvements, because the hard part of an AI agent in finance is control, not capability. Start with one vendor group that sends clean, consistent invoices. Keep the review step on. Give the agent only the permissions it needs. Measure how often a person has to correct its draft. Expand only when that number is low and stable.

That's the real lesson of every AI feature in Business Central. Agents work on the data and processes you already have. If your vendor list has duplicates or your purchase orders are entered after the invoice arrives, the agent will reproduce those problems faster.

On cost: before you budget, check how Microsoft currently bills for agents in your tenant. Terms for AI agents are still changing, so don't plan around a figure from a blog post, including this one.

Wait on these: preview features

Several headline items in version 29 are in public preview, not general availability:

  • The new Microsoft Copilot experience in Business Central. It replaces the older Chat with Copilot feature, but it's in preview, and Microsoft notes that Business Central agents aren't available from it. Access to your wider work data through it also requires a separate Microsoft Copilot license. (Microsoft's documentation.)
  • The Expense Agent updates, including AI-driven approvals and automatic trip-based expense reports.
  • Several tax and excise items for specific countries.

Preview features are fine to try in a sandbox. We don't recommend running finance processes on them in production until they reach general availability. Preview behavior can change, and support is limited.

What you don't need a consultant for

To be straightforward: much of this needs no outside help. The report audit logging is automatic. Batch totals appear by themselves. Tracing G/L accounts and combining reports into one PDF are features your team can learn in an hour.

Where help pays off is in the decisions with risk attached: designing approval workflows that match your actual delegation of authority, cleaning up master data before an agent touches it, and setting agent permissions so a person stays accountable for what gets posted. And if your system has bigger problems, such as slow month-end close, reports nobody trusts or integrations that fail quietly, fix those before adding AI on top.

Know which version 29 features are worth it for you

Get a plain-language view of what to turn on in your system and what needs groundwork first. Start with an AI readiness review, or talk it through in a 30-minute call.