- Feature
- Run data queries with MCP Server
- Availability
- Generally available (October 2026); the Data Query Tools setting is marked preview in the MCP configuration page
- Industry focus
- Food and beverage manufacturing
Food and beverage manufacturers ask their ERP difficult questions, often under time pressure. Which customers received product made from this ingredient lot? Which lines had the most waste last month? What share of our volume goes to retailers with tight shelf-life terms? Business Central 29.0 gives AI agents a new way to answer questions like these, even when no ready-made API exists.
The industry challenge
The Model Context Protocol (MCP) server for Business Central lets AI agents and assistants connect to ERP data in a standard way. Until now, those agents could only use data exposed through APIs, meaning API pages and API queries configured for the MCP server.
That works well for everyday entities like customers, items and sales orders. It works less well for the questions that matter most in food and beverage:
- Joining item ledger entries, lot numbers and customer shipments for a traceability check
- Aggregating output and scrap by production line and week
- Comparing purchase prices for a commodity ingredient across suppliers and periods
Each of those needs a join or aggregate that no standard API covers. Building a custom API for every question is slow and expensive, so the questions go to a report writer, or to Excel.
What Microsoft delivered in Business Central 29.0
Business Central 29.0 adds MCP tools that let an MCP application define, validate and run its own data queries. This means it can query data even where no API exists.
According to an MVP walk-through of the release, the new Data Query Tools setting on the MCP Server Configuration page adds four system tools. They let an agent:
- Find tables by name or by meaning
- Look up how a table relates to other tables
- Read the fields in a table
- Compile and validate an AL query, and optionally return the data rows
The tools are read-only. Agents can't create, change or delete data through them. API pages and queries you have already exposed keep working independently.
Where the value shows up
| Business area | Impact for a food and beverage manufacturer |
|---|---|
| Traceability | Faster answers to "where did this lot go?" questions |
| Operations | Ad-hoc analysis of yield, waste and output without a report project |
| Purchasing | Quick comparisons of commodity costs over time and by supplier |
| IT backlog | Fewer one-off API or report requests |
| Data safety | Read-only access keeps exploration away from transaction data changes |
The read-only design is the reason this is safe to try. Exploratory questions are exactly where you want AI help, and exactly where you don't want an agent to change anything.
A worked scenario
Consider a bakery manufacturer that supplies supermarkets and food-service customers (an illustrative example). A flour supplier warns that one delivery may be contaminated. The quality manager needs to know, quickly, which finished-goods lots used that flour and which customers received them.
In the past, the IT lead would build an ad-hoc report or export ledger data to Excel and join it by hand. That took hours on a good day.
With data query tools enabled on the MCP server, an authorized assistant can find the relevant tables and understand how they relate. It then builds a query that follows the lot from receipt through production to shipment. The quality manager reviews the result and checks it against the item tracing page in Business Central. Then they decide on the next steps, with time to spare.
Adoption considerations
- Treat it as preview where marked. Check the status in your MCP configuration and pilot before relying on it for critical decisions.
- Control who can use it. MCP access follows Business Central permissions and your MCP configuration. Decide which users and agents should get the query tools.
- Validate critical answers. For recalls and regulatory questions, confirm AI-generated results against the standard traceability features.
- Promote repeat questions to proper reports. If the same query runs every week, turn it into a report or API for stability.
- Mind data sensitivity. Customer and pricing data are commercially sensitive. Apply the same governance you would to any analytics tool.
The bottom line
Food and beverage manufacturers live with tight margins, short shelf life and strict traceability rules. They need answers faster than a report project can deliver. MCP data queries in Business Central 29.0 let AI agents answer questions that no API was designed for, safely and without writing to your data. Used with sensible governance, this can turn the ERP into a source of quick, trusted answers.
If you're exploring how AI agents could work with your Business Central data, it's worth planning your MCP configuration with an experienced adviser.
Sources: Microsoft roadmap 573312 · Dynamics 365 Lab: Run data queries with MCP Server