Microsoft now ships a native Business Central MCP server, so AI assistants can reach Business Central directly. The hard part is not connecting them - it is doing it safely: which data, read or write, for whom, and with what audit trail. That governance layer is where TrustFort works.

We do not build a Business Central MCP server from scratch. We configure, govern, and extend Microsoft's - and build custom MCP servers for everything it does not cover: SharePoint, Confluence and Jira, your SOPs and validation documents, and cross-system agent workflows. In a regulated environment, permission design, read-versus-write scoping, auditability, and de-personalization to service accounts are exactly the parts that decide whether AI access is defensible.

Three layers

Business Central MCP

Enable, govern, and extend Microsoft's native Business Central MCP server for Copilot Studio and Claude / ChatGPT clients.

Company knowledge and RAG

Make SOPs, validation documents, and project history queryable by AI - a governed retrieval layer over your own content.

Custom MCP servers and governance

For systems without a native MCP server, and for cross-system agent workflows - built with the governance thread throughout.

Business Central MCP:
- Enablement and API-page configuration; client setup for Copilot Studio, Claude, and ChatGPT
- Permission and governance design - read vs. write scoping, least privilege, auditability
- Extension beyond the read-only defaults where the use case needs it, under control
Company knowledge and RAG:
- Retrieval-augmented generation over your documents on Azure AI Search + Cosmos DB
- SOPs, validation records, and project history answerable with citations, not guesses
- We run this exact pattern in production for our own company knowledge - the approach is proven, not theoretical
Custom MCP servers and governance:
- Custom MCP servers for SharePoint, Confluence, Jira, and line-of-business systems
- Cross-system agent workflows that read and act across more than one system
- Governance by design - permissions, read/write scoping, audit trails, de-personalization to service accounts, tied to your GxP controls

How we deliver

Governance-first: every integration starts from what the AI may see and do, not from what is technically possible.
Time-boxed engagements with a written deliverable - the configured or built integration, its permission and audit design, and a handover.
Senior involvement throughout. The people who design the access model are the people who build it.

Give AI safe access to your Microsoft stack

If you are enabling Business Central MCP, planning a company-knowledge assistant, or need AI to work across systems without losing control of permissions and audit, we offer a focused discussion to clarify scope and next steps.

Discuss AI integration

Related service: BI & Data Platform for Business Central.