Colin Frishberg

I build AI products for people who have to make decisions with them.

I’m an AI Product Lead at Google, where I’ve spent the last five years building the sales platforms and reporting infrastructure our global Ads organization runs on. I’m currently focused on democratizing data analysis through natural language queries across our complex Business Intelligence suite.

In parallel I’ve just completed a Master’s in Information & Data Science at UC Berkeley, focused on Generative AI and Causal Inference. That’s been deliberate: I wanted to be able to build and rigorously evaluate the AI systems I’ve been shipping in industry, rather than take demo quality on faith. My recent work includes Vantage AI, a production-grade lease analysis web and Android application built on a hybrid deterministic and LLM pipeline with retrieval-augmented generation and agentic loops, and a causal study using regression discontinuity in time to measure the impact of automated speed enforcement.

My first four years were in enterprise software sales — CIAM at Gigya, UCaaS at Arkadin, IT research and advisory at Gartner. It’s where I learned to hear what a user actually needs underneath what they’re asking for. I went to Berkeley for the other half of the job: enough depth to spec an eval, read the results, and push back when a demo is outrunning the data.

The short version: I’ve sold enterprise software, built it, and learned to measure it. I’m most useful where those overlap.

Before all of it: an undergrad at the University of Michigan — Berkeley makes it two schools in blue and gold. Go Blue, and Go Bears.

Outside work: motorcycle racing, causal inference, and the political economy of everything.