The situation
A vertical SaaS company serving mid-market healthcare providers was leaking renewals. The signals were there — usage dips, support ticket patterns, admin changes — but they were scattered across three systems that only a motivated rep would ever correlate. RevOps leadership had modeled churn for two years without ever getting a risk signal into the hands of the person who could act on it.
What we shipped
A two-person Vert3x team — a delivery lead and an ML engineer — embedded with RevOps and CS for six weeks.
- A renewal-risk model fusing product telemetry, support signal, and CRM activity into a single per-account risk score, refreshed daily.
- An agent layer that drafts the outreach play (which signal fired, which motion to run, who on the account to contact) and routes it into the CSM's queue inside the existing CRM.
- A pipeline-grooming agent that keeps the CRM honest — flags stale stages, missing next steps, and suspicious close dates — so the forecast review stops being a forensic exercise.
- A transfer document and a monthly operating rhythm that the client's RevOps team now runs on their own.
Results
The value isn't that the model is smarter than our CSMs. It's that the signal gets to them on the day it matters instead of on the day we'd normally notice. — CRO
- Renewal-risk accounts flagged ~34% earlier than the prior manual process.
- ~11 hours per rep per week reclaimed from admin and pipeline hygiene.
- Six weeks from kickoff to a live production system — including the agent review UX.
- A renewal-risk review cadence the CRO now runs weekly, without any Vert3x presence in the room.
Why it worked
We didn't rebuild their CRM. We surgically added one model, one agent, and one UX surface — inside the system their team already lived in. The fastest way to an AI-native RevOps function isn't a new platform. It's one operating loop, made faster by the thing you already have.