AI Data Wizard: Making Business Central Implementations Faster and Reliable

Every Business Central implementation starts the same way: turning a customer's raw data — spreadsheets, item lists, price files — into dozens of correctly linked master records before go-live. This work is slow, repetitive, and error-prone, and because every customer's data is structured differently, a single fixed template rarely fits. This session introduces the AI Data Wizard, a proof of concept in which an AI agent, Claude, interprets whatever raw data a customer provides, maps it to the target data model, and drives Microsoft's official Business Central MCP server to create the records — with a consultant reviewing and approving every change before it commits.
Because the mapping isn't hardcoded, the wizard adapts to each customer's data instead of forcing data to fit it, so implementations get configured faster and trusted more confidently. The session covers how the MCP server's write-safety model works in practice — write access enabled per API page by a BC admin, environment and company chosen via configuration headers, every operation run under the consultant's own identity — plus the safeguards layered on top: sandbox validation before any live use, a block on writing to a live company without deliberate authorisation, and a clearly scoped agent.
Attendees leave with a working example of how AI-assisted data onboarding can cut implementation time without compromising control — what worked, what didn't, and where human oversight still matters.