VC Automation - AI research and outreach for fundraising

Partners at VC firms get hundreds of cold emails a week. The ones that get a reply open with something the partner actually said. VC Automation finds the right partner at the right firm and says something they care about this week.

It tracks what partners publish: a podcast episode, a Substack post, a conference talk. Enrichment pulls the transcript or full text and extracts what the partner argued for. Scoring weighs thesis fit and recency. Anything under the threshold is dropped, which keeps the volume honest and the inbox reputation clean.

The message is three sentences. The first refers to the argument the partner made. The second ties it to the founder's thesis. The third asks for fifteen minutes. No deck, no long pitch.

The model never runs the tools itself. It reads a deal and writes a plan, and ordinary code runs the plan with retries and error handling. That one boundary took the system from 89% reliable in prototyping to 99.6% in production.

Role: solo build

Partner content as signals, scored for thesis fit and recency

Model plans, code executes

89% to 99.6% reliability from prototype to production