Inside Startup Due Dil: the AI tool turning hours of VC diligence into 10 minutes
Tech.eu Cate Lawrence
A VC partner built an AI tool that turns startup due diligence into a 10-minute first pass. It’s cheap, fast, and still meant to help humans, not replace them.
Based on reporting by Tech.eu, Cate Lawrence — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Vlad Tislenko, a partner at SMRK VC, has launched Startup Due Dil, a web app that automates the messy part of startup due diligence: collecting, checking, and organizing information. SMRK VC backs Ukrainian tech startups, and Tislenko says he built the tool first for his own work as an investor. He even says the internal version cost less than €1,000 to make.
The pitch is simple. A process that can take several hours or even several days is trimmed into an initial structured report in about 10 minutes. Startup Due Dil starts with a pitch deck upload, then can pull in financial statements, cap table data, legal documents, written comments, and links to other sources. It compares the claims in those materials with public information and marks findings with red, yellow, and green flags.
Under the hood, the system runs ten AI agents. Nine of them focus on specific areas: team, market, competitors, technology and product, traction and financials, business model, regulatory compliance, ownership structure and cap table, and legal issues. The tenth, called Oracle, coordinates the whole thing, checks whether the work is complete, and can send agents back for another pass if something looks thin.
Tislenko is candid about what pushed him to build it. He says he wanted to keep up with AI progress and use it to make better investment calls in AI itself, while also helping portfolio companies. He also says the tool has repeatedly exposed a familiar startup habit: founders oversize their market numbers, and those figures often cannot be verified.
That said, this is not an argument for letting software run the fund. Tislenko says due diligence still depends on talking to people, doing reference checks, and working through signals with founders. Privacy is another issue, so the platform avoids exposing private workspace content in admin dashboards, and developers do not have access to user files or reports. The current version uses the OpenAI API, though it can also support Anthropic, and Tislenko says he designed it so neither model provider can use customer materials to train their systems.
My take — AI-written commentary, not fact-checked reporting
This is the sensible AI bet: not “replace the investor,” just kill the dullest 80% of the slog. The industry keeps selling magic, but the real prize is a faster first pass with citations and fewer heroic guesses. That’s not flashy; it’s better.
Read more about this at: Tech.eu
