Case study · Consumer finance, United States

From zero to 28 branches live, averaging a new one every three days.

A lender with branches had a list of past customers nobody had time to call. We built an AI voice layer on Atlas that qualifies, books, writes the CRM notes and follows up. And we answered for the number, not for the dashboard.

28 branches live, reported countBuilt on AtlasClient name withheldDelivered through PTS AIFigures as reported by the account's platform and CRM; period not stated per figure

Where it started0No system.

→

Where it ended28Branches live, with the same system.

The problem

The customers were already there. Nobody had the hours to call them.

A lender's branches know exactly who paid off a loan. Nobody calls them back, because every hour goes to the customers who walk in. There is more list than hours. And across many branches, nobody can see which one is leaving money on the table.

What was built

Not a bot. The first mile, written down and automated.

The call layer.AI calls that work the list, qualify, and pass a live person to a branch rep.

Every outcome in the CRM.So nobody has to remember what happened.

Booking and follow-up.Meetings booked without a person in the middle.

Reporting.Built on the Atlas API, so "which branch broke this week" has an answer on Monday.

Self-hosted automation.On the company's own server, so cost stopped growing with volume.

What it produced

The numbers, and what each one means.

~29,500dialsThe volume no team that size reaches by hand.

76%connect rateMeasured, not estimated.

346live transfersHanded to a human while the person was on the line.

93sales handoffsTraced from first dial to the rep.

33meetings booked

$193Kqualified pipelineWith 5 signed deals behind it.

14.2K+leads processedEach outcome in the CRM.

14live dashboards

28branches live

Time to live

The number we actually work on.

What matters is how long one branch takes to go live. That time went down during the rollout, because each new branch reuses what the first one taught us.

Status

Reported as running at the last published cut. Not re-verified since.

The numbers here are a cut, not an ending. What changes every month is who gets called, at what hour, and what counts as qualified. Those decisions are the conversion rate. We build it with Atlas, and where a platform does it better, we use the platform.

What this does not prove

The honest edges.

One account, one industry.
Long enough to show pipeline and signed deals, not retention.
The client's name is withheld until they approve it in writing.
Nothing was rounded up for this page.

Start working with us.

Tell us what is breaking and we will tell you straight whether this is even a hire. You keep the map of your own process either way. Free, and nobody calls you unless you ask.

Start working with us →