Algorithms Are Deciding Which Startups Get Seen By Venture Capitalists or Not

Venture capitalists are not only at the forefront of funding transformative artificial intelligence (AI) startups, but AI is also transforming venture capital firms themselves.

A new study from Saïd Business School at the University of Oxford finds that fewer start-ups receive direct partner attention, highlighting the scale to which AI is increasingly used to filter entrepreneurial opportunities before any human review. Thomas Hellmann, researcher and professor at Saïd Business School, says, “AI systems have effectively become a gatekeeping infrastructure between sourced deals and partner attention, fundamentally changing how investment pipelines operate.”

The research is based on qualitative analysis of venture capital firms, including interviews with investors and case studies of firms using AI in their workflows. It examines how these tools are applied in practice, particularly in sourcing and screening deals, to understand their real impact on decision-making.

The study finds that AI is actively determining which companies are even considered by investors. Screening models apply explicit thresholds to determine which companies advance to human review, creating a system in which early-stage decisions are increasingly automated.

These systems process vast volumes of startup data. One firm’s AI system reviews around 1,000 startup leads per week, narrowing them to just a few dozen opportunities.

This shift is already influencing founder behaviour. Entrepreneurs are beginning to optimise their profiles and materials for algorithmic systems, with some fine-tuning websites and profiles to match the keywords and categories they believe AI-based systems prioritise.

The research also raises significant concerns about bias embedded within these systems. “Models trained on historical venture data may favour founders who resemble those previously backed,” says Hellmann, “reinforcing existing patterns in investment decisions.”

Hellmann warns that, if left unchecked, these systems risk hardwiring past preferences into future portfolios and systematically excluding atypical founders. “And this is an industry that already has a poor track record on diversity ”

The findings suggest AI may reshape the types of companies that receive funding. Evidence cited in the study shows that AI-driven investment strategies tend to favour startups that resemble past successes, improving predictability but reducing support for more radical ideas.

This raises concerns that the industry could become less effective at identifying outliers, the high-risk, high-reward ventures that drive transformative innovation.

Despite widespread adoption of AI systems, the study finds that final investment decisions firmly remain in human hands. AI tools are primarily used to structure information and support analysis, but the act of deciding and owning consequences remains with humans.

“Yet we are beginning to see some cracks appearing there too,” adds Hellmann. “Some venture capitalists are beginning to use AI in Investment Committees, where investment decisions are made. There, venture capitalists may question the AI agent, ask them to play devil’s advocate, or request a consultative vote. Still, the AI agents can’t formally vote: after all, that is what the venture capitalists get paid for.”

Overall, the research highlights a fundamental transformation in how venture capital operates. AI is expanding the number of startups investors can track, but simultaneously limiting the subset that receives meaningful attention.

AI is redefining access to capital, determining not just which startups succeed, but which are even seen.

ENDS

If you’d like to see a copy or know more about the research paper or speak with Thomas Hellmann, please contact adam@bluesky-pr.com