Grow Therapy
The systems that let other people use AI well.
When a product team needs AI inside a real workflow, give it reliable company context, governed data and tools, evaluations that say whether the output is good, and measurement around the product that results.
My work at Grow Therapy has shifted from doing data science myself to building the systems other teams use to work with AI. Getting a model to answer is rarely the hard part. The hard part is everything around it, and that work falls into the three areas below.
Case studies
- 1.
AI Enablement Platform
Let other people inside the company build on AI without each team inventing context, tools, and evals from scratch.
Built provider intelligence from fragmented sources (structured LLM extraction and identity resolution), governed semantic-layer context connected to Claude through MCP, evaluations for AI-generated analysis, and reusable AI workflow tooling. The aim is to turn one-off model use into an internal system the rest of the company can safely build on.
- 2.
Experimentation & Causal Measurement
Know whether a product change worked after the measurement stack itself has moved.
Set up Statsig Warehouse Native, and built the identity resolution, validation, rollout measurement, and causal-inference layer around it.
- 3.
Marketplace Decision Systems
Decide who to surface, how to match supply and demand, and what a provider or price move is worth.
Ranking, forecasting, provider quality, supply and demand, LTV, and elasticity models. A longer-running body of marketplace modeling, not one bounded project.