An owner-led commercial insurance brokerage was losing nearly four in ten business-hours callers to rings that ended in nothing, and at lunch nobody answered live at all. We read every call leg, found the causes, rebuilt the answer chain with an AI assistant behind the people instead of in front of them, and measured the result with one engine before and after.
What we built, and what it moved.
Client names withheld by choice. Every number on these pages traces to a dated read of the system it came from.
- Commercial insurance brokerage Insurance · Phones + AI assistant38% → 13% Callers who reached nobody Main line, weekday business hours · June–July vs the four weeks from Aug 17 · one engine, both sides
- Storm-restoration contractor Roofing · CRM automation10% → 80% Signed jobs carrying a contract value 3 of 29 on Jul 2 → 53 of 66 on Sep 14 · values entered by the contractor's own team
A storm-restoration contractor's CRM held its signed work, but only 3 of 29 signed jobs carried a dollar value, so the pipeline could not be read in money. We built an automated operating report that has landed on phones at 8 a.m. for 75 mornings running, flagged the gap, and fixed a stale sync of our own that had understated collections by $338,531. Today the report tracks $1.25M in contract value across 53 of 66 signed jobs.
- No Brainer Media Our own operations314 Changes merged in under seven weeks One repository, Jul 30 → Sep 14 · every merge passed the required test gate
The client work in these case studies runs on an operating system we built for ourselves first: a knowledge graph that remembers every decision, AI engineering sessions that ship through tests and pull requests, scheduled readers that take numbers from the source, and gates that stop a change before it reaches a client. This is what it is, measured, and what it has caught.
The next case study might be yours.
Book a discovery call. We will tell you whether the work you need is what we ship — or not.