Angel Lens

A web app that turns a start-up’s pitch deck into a structured analysis in the style of an institutional investment committee, built for angel investors.

Question
How close can an angel investor get to the discipline of an institutional fund’s investment memo, starting from a single pitch deck?
Status
Prototype
Started
March 2026
Tools
  • Claude API
  • React
  • PDF.js
  • Vercel

The problem

Angel investors rarely have the analytical capacity of an institutional fund. I wanted a tool that applies the evaluation discipline I used for years in venture capital and turns a pitch deck into a structured investment analysis.

What I built

A web app that takes a start-up’s deck and returns an 18-section analysis:

  • venture capital evaluation criteria and a scorecard,
  • market size (TAM/SAM/SOM), competition and comparable companies,
  • go-to-market strategy, team and defensibility,
  • financial profile, valuation and return scenarios, exit analysis,
  • a risk matrix, questions for the founder meeting, an investment memo and a one-page summary.

The interface works in Turkish and English, and results can be exported to PDF and PowerPoint.

What worked

  • Extracting text from large PDFs in the browser, instead of uploading the files, cut the payload from megabytes to kilobytes.
  • Anchoring the framework in a real investment memo and criteria set made the output far more consistent.

Angel Lens is a decision-support tool; it does not produce investment advice.

All experiments

View all
Experiment 01Prototype

CourtScribe

The court’s scorekeeper; an AI system that watches game video and keeps the player-by-player statistics table itself.

Can player-level statistics be extracted from an amateur basketball game video without anyone keeping score by hand?

  • YOLOv8
  • ByteTrack
  • EasyOCR
  • faster-whisper
  • Google Colab

June 2026

Experiment 03Prototype

VisemIQ

An experiment in turning speech into text from lip movements alone, in video without sound.

Can speech be read from video alone, without any audio?

  • Visual speech recognition
  • Large language models
  • Python
  • Claude

September 2026