Dealflow: Building a Smart Deal Screening Engine at Dark Alpha Capital

Dealflow: My Centralized Deal Screening Engine at Dark Alpha Capital

At Dark Alpha Capital, our independent sponsor private equity firm, deal flow never stops. Leads roll in every day from the go-to-market and marketing teams. For a long time we handled them the old-fashioned way — scattered emails, spreadsheets, and manual reviews. It worked, but it was slow and easy to miss things. So I built Dealflow to change that. It’s a single system that takes every raw lead, screens it properly, and keeps everything organized in one place.

The process runs in two clear stages.

First comes the quick, deterministic pass. The system automatically checks every lead against our hard requirements: minimum EBITDA, minimum revenue, whether it fits our target industries, and if the deal location makes sense for us. Anything that doesn’t pass gets flagged right away. No guesswork, just clean filters that cut out the noise fast.

After that, we move to the deeper screening. We have a set of internal screeners — the key questions we always ask about any opportunity. For each lead, Dealflow stores the full set of documents, financial history, contact details, and notes. I embed everything with Cloudflare Vectorize so the system can search and reference the actual content. When we run a screener, AI pulls the relevant answers and points back to the source material. It’s like having an always-on research assistant that actually knows the files.

I also added a tool called Chatflow because we needed something even simpler for the team. Associates can now just type natural questions — “What deals came in this week?” or “Show me healthcare opportunities above $5M EBITDA” — and get clear, structured answers back. It runs on tool calling, so the responses stay accurate and actionable instead of hallucinated.

On the tech side, I leaned hard into Cloudflare. Vectorize handles all the embeddings and similarity search. Cloudflare Workflows manage the background jobs that process leads step by step without breaking. The whole app runs on Bun, uses Drizzle as the ORM with PostgreSQL for the database, and TanStack Query for clean frontend data fetching. I originally built it in Next.js, but I moved it to TanStack Start because the deployment experience on Cloudflare Pages is just smoother for this kind of full-stack setup.

We also keep investor leads and high-net-worth individuals in the same system. That way everything — deals and capital — lives together when we’re raising for a transaction.

Dealflow isn’t flashy, but it actually makes our day-to-day work faster and more consistent. What used to take hours of digging now happens in minutes. It’s still evolving, but it already feels like the backbone of how we run deal screening at Dark Alpha Capital.