Vlad Tislenko, a partner at Ukrainian venture capital firm SMRK VC, has launched Startup Due Dil, an AI-powered platform designed to automate key parts of startup due diligence. The system uses ten coordinated AI agents to collect, verify, and structure information provided by startups alongside data available from public sources. Originally developed as an internal tool, the platform aims to reduce the time investors spend on routine research while providing a structured evidence base for investment decisions.
A Multi-Agent Approach to Due Diligence
Startup Due Dil divides the assessment process among nine specialist AI agents, each responsible for a specific area of startup analysis. These areas include founders and teams, markets, competitors, technology and products, financial performance, business models, regulatory compliance, ownership structures, cap tables, and legal considerations. A tenth agent, known as Oracle, coordinates the process, evaluates the specialists' work, and compiles their findings into a final report.
Users can begin an assessment by uploading a startup's pitch deck, while additional documents such as financial statements, cap tables, and legal materials can be supplied for deeper analysis. The platform compares information contained in these documents with publicly available sources and organizes the resulting findings according to their significance. Reports use red, yellow, and green flags to distinguish potential risks and inconsistencies, issues requiring further clarification, and positive findings that have been verified.
Reducing Manual Research
Tislenko initially developed Startup Due Dil as a personal project to improve his investment analysis and deepen his understanding of emerging AI capabilities. The system was built for less than €1,000 and is now being used within SMRK VC's investment process while also being tested by other investors. Its automated workflow can generate an initial structured due diligence report in approximately ten minutes, compared with the hours or days that more traditional research can require.
One recurring issue identified through the platform is the difficulty of independently confirming market-size estimates presented by early-stage startups. Founders can rely on ambitious assumptions when describing their addressable markets, making external verification an important part of evaluating investment opportunities. Tislenko has continued refining individual agents when weaknesses emerge, using feedback and repeated testing to improve how the system identifies nuances and inconsistencies.
Managing AI and Data Risks
The use of generative AI in investment analysis introduces its own risks, particularly because models can produce convincing conclusions that are inaccurate or insufficiently supported. Startup Due Dil's multi-agent architecture attempts to reduce this problem by having Oracle assess the quality and completeness of specialist outputs and rerun individual agents when additional work is required. The system is nevertheless intended to support human analysis rather than eliminate the possibility of AI errors.
Data privacy is another important consideration because due diligence can involve confidential financial, legal, ownership, and fundraising information. Private workspace content is separated from administrative access, while uploaded materials are temporarily processed through third-party AI infrastructure to produce the analysis. The platform currently uses OpenAI's API and has also been designed to support Anthropic models, with safeguards intended to prevent customer materials from being used to train those providers' models.
Supporting Investors and Founders
Startup Due Dil is primarily designed for venture capital and private equity investors assessing technology businesses, particularly companies at earlier stages of development. Startups can also use the platform independently before fundraising to identify weaknesses in their materials and evaluate their preparedness for investor scrutiny. This could make automated diligence useful on both sides of the fundraising process, helping investors investigate opportunities while giving founders greater visibility into potential concerns.
Despite the automation, the platform does not attempt to replace direct conversations, reference checks, or the judgment investors develop through experience. Human interaction remains particularly important when findings require context or when investors need to evaluate founders, teams, and relationships that cannot be fully captured through documents and public information. Instead, Startup Due Dil positions AI as a tool for handling research-intensive tasks and creating a more systematic foundation for further investigation.
Startup Due Dil reflects a broader push to apply multi-agent AI systems to professional workflows that traditionally require substantial manual research. By coordinating specialized agents and combining startup-provided materials with external information, the platform seeks to accelerate initial due diligence without removing investors from the decision-making process. Its development also highlights the continuing challenge for AI-powered investment tools: balancing speed and automation with verification, confidentiality, and human judgment.