AI Company Analysis Framework

An async LLM pipeline that turns ~120 minutes of manual company research into a structured, sourced report in ~35 minutes. Runs free on Gemini/Groq free-tier models, or ~$0.005–0.02/run on OpenAI. Used to analyze 70+ companies.

Full reports the tool generated:

• → BlackRock (BLK) — generated free on Gemini, $0.00
• → Apple (AAPL)

How it works

COLLECT ───▶ PREPROCESS ───▶ ANALYZE ───▶ REPORT
financial     clean +          5 modules:   Markdown
(yfinance),   deduplicate +    financial,   + JSON
news (RSS),   token-optimized  sentiment,   with sources
competitors   chunking         competitive,
(concurrent)                   risk, growth

• Concurrent collection — financial data, news, and competitor info gathered in parallel
• Token-optimized preprocessing — dedup, relevance filtering, chunking
• 5 analysis modules — financial health, news sentiment, competitive position, risk, growth
• Rate-limited async API — token bucket + exponential backoff
• Structured output — Markdown reports + JSON, with source metadata

Code, tests, and setup: github.com/changsunglim/ai-company-analysis-framework