BullBrief

Equity research starts the same way every time: pull the filings, skim the earnings call, check the fundamentals, and write it up. BullBrief does the pulling and the first draft — and it's the project I've kept rebuilding the longest, from an LSU scraper into the platform running at bullbrief.pro today.

The frontend and backend split into their own repos once the original monorepo outgrew it: bullbrief-frontend deploys to Vercel, bullbrief-backend deploys to Render behind a Render healthcheck route and an explicit CORS allowlist scoped to bullbrief.pro and local dev. The API itself is 20+ Flask blueprints; the core of it is routes/analyst.py, a ~38KB market-analyst engine that grades sector peers, blends valuation signals, and produces the analyst-style read the frontend renders.

The hard part wasn't the AI — it was orchestrating high-concurrency financial data streaming and multi-source ingestion across five upstream sources that fail independently, and keeping the LLM output honest by validating it against Pydantic schemas before it ever reaches the page.