Case study
An AI agent that chases every estimate
Future Group was spending around 20 hours of management time a week keying estimate emails into a spreadsheet and chasing clients for answers. An agent now reads the inbox, logs every estimate, drafts the chasers and reports the pipeline.
2025–today
- 400+
- estimates capturedone month
- 600+
- emails handled by the agentone month
- 80+ hours
- savedone month
- 9 users
- across 3 divisions, 100% adoptionwithin 4 weeks
The challenge
Estimates went out from three divisions and sat. Someone had to copy each one into a spreadsheet, remember to chase it, dig the quote out, write the email and hope. The person doing it was leaving.
What we did
- May 2025 — scoped to automate estimate tracking across every division, with people keeping control of what goes to clients
- Version 1: the agent extracts estimate details from incoming email, presents them in a review queue, and drafts first chasers for approval
- Version 2, proposed November 2025: full automation — the agent classifies every email, marks estimates won or lost, adjusts chase timing to the client's situation (Section 20 consultations, tender timelines) and only forwards what needs a human
- Analytics, proposed alongside: win rates by managing agent, pipeline value by division and client response times to each chase
What we built
- Inbox capture with no manual data entry
- Review and approval queue
- Automated chase emails
- Full visibility of the estimate pipeline
- Works from the Microsoft 365 estimates inbox
The result
Nine users across three divisions adopted it in four weeks, with no training and no complaints. Spreadsheet tracking stopped, and the role that ran it was almost entirely replaced before the person in it left.
Stack
- Microsoft 365
- AI extraction
- Web review UI