Pediatric Therapy Startup
Midwest Data went from zero to a fully modeled warehouse in three weeks — investors noted that we had a "Series-B architecture in a seed-stage company". The foundation has compounded ever since: reporting that used to take weeks now happens automatically, and they moved at startup speed without cutting any corners.
Context
A seed-stage provider of in-home pediatric therapy — SLP, OT, and PT services delivered in patients’ homes. Under ten people on the team. The data environment was a warehouse in name only: spreadsheets everywhere, a handful of messy Metabase SQL scripts pointed at transactional databases, and no pipelines, no dbt, no modeling layer.
Problem
Two things were breaking at once. Operations were scaling — more providers, more sessions, more payers — and reporting couldn’t keep up. At the same time, the company was preparing to raise a Series A, which meant investors would want clean, defensible metrics that didn’t yet exist. Without a data foundation, both the raise and the growth trajectory were at risk.
Solution
We came in and built the data function end to end.
The first engagement was a warehouse from scratch: Snowflake plus dbt, with proper modeling for parents, dependents, providers, sessions, claims, and payer relationships. First models and reports were running within two to three weeks. As operations scaled, so did the foundation — over 400 dbt models across staging, intermediate, and mart layers, plus seven ETL pipelines pulling from HubSpot, Bubble, Xano, Google Ads, and provider licensing databases, with snapshots, tests, and CI/CD in place.
On top of the foundation, we’ve built AI systems that compound the team’s leverage:
- Automated CPT billing. Claude generates session-level CPT codes from clinician notes every morning, with shadow-mode validation against historical accuracy before production deployment.
- Natural-language analytics. A Slack bot lets non-technical staff query the warehouse in plain English — Claude translates to SQL, runs against Snowflake, returns results in-thread.
- Provider discovery. An AI enrichment pipeline finds and verifies licensed therapists at $5–7 per 100 providers vs. $500+ from legacy vendors — a roughly 70–100× cost reduction.
The engagement started solo and grew to three engineers as the business scaled.
Outcome
The company closed a Series A. The co-founder credits the data foundation and our execution speed directly: We wouldn't have closed the round without this infrastructure and the speed at which it was built. Investors described the result as enterprise-grade infrastructure delivered at startup speed.
The team has since grown substantially, expanded across multiple states, and runs near-fully automated reporting without proportional increases in data spend. The engagement is ongoing — the foundation keeps compounding.