SMRK VC launches AI tool Startup Due Dil

Venture capital firm SMRK VC has launched a new AI-powered due diligence platform called Startup Due Dil. Vlad Tislenko, a partner at the firm, announced the web application, which is designed to automate the collection, verification, and structuring of information for startup investment reviews.

Tislenko told the source that he built the tool initially for his own work. The goal was to keep pace with AI advancements to make better investment decisions. The platform cost less than €1,000 to develop. It aims to tackle the most time-consuming part of systematic startup analysis: finding, verifying, and structuring information.

A ten-agent system for automated analysis

The system orchestrates ten AI agents to analyze materials like pitch decks, financial statements, and public sources. Nine specialist agents each handle a distinct area of analysis. A tenth agent, called Oracle, acts as the orchestrator for the entire process.

Oracle plans the diligence process, provides context to the specialists, and evaluates their work. If an analysis is insufficient, Oracle can rerun the relevant agent with more context. The final report organises key findings into red, yellow, and green flags, complete with source citations.

Verifying claims and managing AI risks

A user starts an analysis by uploading a startup's pitch deck. For a deeper review, they can add financials, legal documents, and other sources. The platform then compares claims in the submitted materials with public information. According to Tislenko, the tool frequently cannot verify the market size numbers that many early-stage founders provide.

He acknowledges a fundamental challenge: AI systems can confidently produce incorrect information. To minimise this risk, he built the multi-agent system with Oracle setting high standards. The agents work hard throughout the approximate 10-minute analysis window. Tislenko admits there is still a chance the system may make a mistake.

Data privacy is a core priority. Administrative dashboards do not show private workspace content, and developers lack access to user files. The current version uses the OpenAI API, though the architecture also supports Anthropic. Tislenko stated he designed the system so that OpenAI or Anthropic do not use customer materials to train their models. He also claims he does not have access to user inputs and outputs for training, only using logs for customer support.

A tool for investors and founders

Startup Due Dil is designed primarily for venture capital and private equity firms evaluating tech companies, especially at early stages. It is currently used in SMRK's investment work and is being tested by other investors. Startups can also run a one-off analysis before fundraising to assess their readiness for investor scrutiny.

Tislenko emphasised that the tool is not intended to replace the investor. The most important part of a VC's job remains building relationships. Due diligence still requires talking to references and discussing signals with entrepreneurs. He believes that as AI becomes more common in due diligence, it could change how startups build pitch decks and data rooms. These systems might one day help founders build perfect materials. But ultimately, he contends, it will still be a people’s game.