2026 · Web app
AI Viability Scorecard
Can AI help an investment firm research companies and assess what makes them defensible and AI-forward?
The pipeline. Every company gets a verdict from an 8-item rubric: Likely Pass, Watchlist, Deeper Diligence or Prioritize. Dealbreakers show up in red no matter how good the score looks.
I delivered a prototype for a local VC firm who is determining the best way to quickly sort real AI companies from “thin wrapper” companies.
Plenty of AI startups are a thin layer on someone else’s model, the kind of thing any engineer could rebuild in a weekend. I built this scorecard for a private investment firm to spot those early.
Each company is scored 0-3 on eight questions across team, moat, technical depth, economics and a reality check. Written anchors spell out what each score means. The percentage maps to a verdict, from Likely Pass to Prioritize. Three dealbreakers get flagged separately, no matter what the score says.
Research with Claude does the first pass. It reads the company’s own site first (lots of startups share a name), then searches the web and suggests a score for each item with sources. When public info can’t support a score, it marks the item N/A instead of guessing.
Built with Next.js, TypeScript, Tailwind and the Claude API with web search.