- Feature
- Configure AL MCP workspaces dynamically
- Availability
- Generally available (October 2026)
- Industry focus
- IT services firms and Microsoft Business Central partners
Business Central partners and IT services firms work across many customers at once. Their developers switch between projects several times a day, and more of that work now involves AI coding agents. Business Central 29.0 makes the AL tooling behind those agents much more flexible.
The industry challenge
The AL MCP server lets AI coding agents work with AL projects. It gives them access to project-aware tools for understanding, building and changing Business Central extensions.
In a partner business, though, the work is fragmented:
- A developer might fix a bug for one customer in the morning and build a feature for another in the afternoon
- Shared automation, such as nightly checks or code reviews, may run across many repositories
- Agents running unattended can't stop to answer authentication prompts
If each AI session needs its own server process set up in advance for a specific project, the tooling gets in the way. Teams end up maintaining many configurations, or they don't use agents for multi-project work at all.
What Microsoft delivered in Business Central 29.0
According to Microsoft, Business Central 29.0 lets you:
- Start the AL MCP server without a predefined project
- Add AL projects at runtime, as work begins
- Use configured connection details in headless scenarios, without authentication prompts
Microsoft says this lets AI agent hosts start one reusable AL MCP server before they know the target project, then load projects as needed. That supports flexible agent sessions, shared tooling configurations and unattended automation, without a separate server process for each preconfigured workspace.
Where the value shows up
| Business area | Impact for a partner or IT services firm |
|---|---|
| Developer productivity | AI agents follow developers across projects without reconfiguration |
| Automation | Unattended jobs can work across many customer repositories |
| Tooling cost | One shared configuration instead of many per-project set-ups |
| Delivery speed | Less set-up time at the start of each piece of work |
| Consistency | The same AI tooling and standards across all customer projects |
For partners, time spent on tooling is time that can't be billed. Anything that lets AI-assisted development fit a multi-client day without friction goes straight to margin.
A worked scenario
Consider a Business Central partner with 12 developers supporting 80 customers (an illustrative example). The firm uses AI coding agents for code reviews, upgrade preparation and routine fixes.
Before 29.0, each agent session was tied to a project configured in advance. Developers kept several configurations and often skipped the agent for small jobs because setting it up took longer than the fix itself. The nightly code-quality run needed a separate set-up for every repository.
With Business Central 29.0, the team runs one AL MCP server configuration. When a developer opens a ticket, the agent loads that customer's project on demand. The nightly automation starts the server once and works through each repository in turn, using stored connection details with no sign-in prompts.
Adoption considerations
- Secure stored credentials. Headless connection details are powerful. Store them in a secrets manager and limit what they can access.
- Separate customer environments. Make sure agents can't mix up projects or environments between customers.
- Standardize the configuration. A shared, version-controlled set-up keeps every developer on the same tooling.
- Log agent activity. Combine this feature with AL MCP file logging to keep a record of what agents did.
- Start with low-risk automation. Code reviews and analysis are good first uses before agents make changes.
The bottom line
Partners and IT services firms are among the first to feel the benefit of AI-assisted development, and among the first to feel its friction. Dynamic AL MCP workspaces in Business Central 29.0 remove one of the biggest sources of that friction. One flexible toolchain can now serve many customers, interactive and unattended alike.
If your firm is building AI into its Business Central delivery model, it's worth reviewing your agent tooling and security set-up as part of the 29.0 rollout.
Source: Microsoft roadmap 573348