What we plugged in
Two of our AI agents, wired into the contractor's existing ServiceM8 diary and the BT landline + mobile call-forwarding chain. No new phone system, no new CRM, no price-list rewrites — we loaded their own price matrix verbatim.
Answered every missed call inside 12 seconds with a real-sounding response. Took the job type (EICR, fault find, fuse board, EV charger, rewire, kitchen sockets), the postcode, and the urgency. Sent a ballpark text quote inside three minutes — pulled from the contractor's own price matrix — and offered three survey slots from the engineer's actual diary. Booked straight into ServiceM8.
Chased every unbooked quote at 24h and 72h with a different angle — offering an earlier slot, or breaking the job into two visits. Handed any quote over £2,000 back to the owner with a one-line summary instead of auto-confirming. Hard rule: nothing commercial above two grand without a human.
The numbers, week-by-week
14-day rolling average — first 90 days vs the prior 90 days. Not a one-off spike.
Callback response time
4h 47m (before)
14s (after)
Missed-call recovery rate
31% (before)
92% (after)
Surveys booked per week
4 / wk (before)
15.8 / wk (after)
Owner evening callback hours
7.5 hrs / wk (before)
1.1 hrs / wk (after)
Plumbed into the tools they already used
- ServiceM8
- WhatsApp Business
- BT call-forwarding
- Stripe (deposits)
- Google Reviews
- Checkatrade
- Xero
No new phone system. No replacement CRM. We sit on top of what's already running on the van's iPad.
About this case study: Composite based on outcomes across 5 similar small UK electrical/trades clients (1–3 vans) over the 90 days to 2026-04-30. Names changed at the operators' request. Underlying numbers anonymised and rounded. Individual results vary — we'd rather a slightly dull true story than an inflated one. More like this: trades sector page · Wandsworth dental story.