Case study · Healthcare

A private spine for patient intelligence.

A regional healthcare provider wanted the upside of AI without renting it — or shipping patient data across borders. This is how we built the thing they now own.

43%

faster claim adjudication

2.4h

saved per clinician, per day

0

data leaves the region

The situation

Three facilities, 40k+ patient records a month, and a claims pipeline running on spreadsheets and manual review. Clinicians were spending hours each day on documentation. The leadership wanted AI assistance — but their compliance team had three hard constraints: no patient data leaving the region, a full audit trail on every model decision, and no dependency on a vendor whose roadmap they couldn't inspect.

The problem

  • · Claims review was manual, slow and error-prone
  • · Clinicians burned ~2.4h/day on documentation
  • · Every AI vendor proposed sending data to their cloud
  • · No audit trail, no regional residency, no ownership

The fix

  • · A private data spine deployed inside their region
  • · 33Street models tuned on de-identified claims data
  • · Full audit telemetry on every model decision
  • · The stack, handed over as owned infrastructure
How we shipped it

Forge to handover, in three moves.

01

Audit

Mapped the claims pipeline, document flow and compliance surface across three facilities.

02

Build

Deployed the 33Street model family on a private data spine with full audit telemetry.

03

Own

Handed the team a governed intelligence layer they tune, run and extend in-house.

"We didn't buy a tool — we took ownership of the intelligence layer itself. For the first time the compliance team wasn't the one saying no; they were the ones who approved it."

Director of Operations · Regional Healthcare Group

Kuala Lumpur · Malaysia

Let's forge something extraordinary.

Bring us a problem. We'll bring the models, the agents and the engineering to make it real — whether it's your product, your brand, or your next era.