The context
A company of Fournier’s size — 1,200+ employees, dozens of job sites, a fleet of 850+ pieces of equipment — regularly produces reports and analyses built on its data. That data lives in several systems: the ERP, the logistics platform, timesheets, fleet management, accounting, SharePoint and email.
The problem
Every report rested on one person compiling the data and building the document — several days of work each time. For a truly critical question, someone would dig and analyze for days. For the rest, the analysis simply didn’t happen: crossing a supplier’s three-year history across every job site took staff the company didn’t have.
What we built
An MCP server — the standard that lets AI agents plug into external systems — connecting Claude to every company data source: the ERP, the logistics platform, timesheets, fleet management, accounting, plus SharePoint and email through the M365 connector. Claude can search and extract data from every platform, with access managed by role.
Two tools, two uses: Claude Cowork for everyday analyses and reports; Claude Code for the deeper ones. Leadership and analysts received dedicated training.
Beyond ad hoc questions, recurring analyses are prepared and automated: project costs, procurement costs, supplier renegotiation files — the price histories and volumes no one had time to assemble before.
The results
- Six-figure annual savings on supplier renegotiations alone
- Reports that took several days now produced in 15 minutes with Claude Cowork
- Analyses that never happened before, now routine
- Claude in the hands of trained leadership and analysts
- Every data source queryable by AI agents, according to each person’s role