Investor's View – AI Is changing the game, but not the rules

14.07.2025 14:30, Rita Longobardi

Artificial intelligence might be powering the next generation of unicorns—but it's also quietly changing how those unicorns get funded. Venture capital has operated on personal connections, experience, and gut feeling for years. Now, some investors are exploring how AI can help them screen, track, and understand deals more efficiently. But how practical is it in a field that still relies so heavily on people? We spoke with TOP 100 jury members Andreas Goeldi (Partner at b2venture) and Michelle Tschumi (Head of Startup Finance at Zürcher Kantonalbank) about how their work is evolving and what role AI tools may play going forward.

Venture capital hasn't changed much in how it operates—but the tools available are beginning to. While most investment decisions still rely on judgment and experience, some investors have started using AI-driven platforms to organize deal flow, spot trends, and assess risk more systematically. 

From 2023 enthusiasm to 2025 pragmatism
Back in 2023, Andreas Goeldi, Partner at b2venture, was vocal about the potential of AI in venture capital. In an article titled "Can AI replace VCs? Three experiments", he outlined how AI could support typical VC workflows—writing investment memos, conducting market analysis, and making introductions. Rather than predicting replacement, his experiments raised practical questions about how far AI could augment the judgment-driven, relationship-heavy nature of early-stage investing.

Andreas, looking back at that 2023 piece, which parts of those early experiments with AI do you still find relevant, and which ones turned out to be hype?
All the specific examples (market research, document drafting, network mining) are still relevant and are things that we do every day now. However, AI models making recommendations for investments are not something that is quite ready for the real world. One reason is that it's very challenging to get all the necessary context into an AI model (e.g. an entire data room, call transcripts, reference information, etc.) and have it process everything with the right prioritization. Current models are not good at providing the holistic perspective that is necessary for good decisions.



Now that the initial hype has settled, what do you see as the most durable and useful AI applications in your VC work today? Could you share a current example or reflection?
Researching markets and current trends has become much more efficient thanks to AI because with the latest tools, it is possible to get a comprehensive report on a market sector and its competitive situation in 10 minutes. That's a real game-changer because you can have a much more informed discussion with founders if you have this kind of market insight and can concentrate on the really interesting questions. Finding promising startups with AI search tools is also very common, and it's also standard to draft documents, such as investment memos, with the help of AI. We use the product made by our portfolio company, TextCortex, very heavily for these tasks because it allows you to define templates and task-specific agents that save a lot of time.

Yet the limits remain clear. Venture capital is still about trust, reputation, and asymmetric information. Would you back a founder solely because of a model's recommendation?
No, but AI recommendations can be a useful input into a decision process. AI models are good at finding facts that are easily overlooked, researching current market information, and processing large amounts of data. They are still very limited in understanding the whole context of an investment opportunity, however.

That tension between efficiency and judgment is also apparent on the financing side. Michelle Tschumi, Head of Start-up Finance at Zürcher Kantonalbank and TOP 100 jury member, remains firm in her belief that venture capital depends heavily on human connections. "For us, venture capital is deeply rooted in human connections. We focus on understanding the team dynamics and trusting our instincts," as she had expressed at the 2024 TOP 100 Investor Panel. "While we do analyze data and leverage machine learning for portfolio insights, our approach remains profoundly personal and relationship-driven."



Michelle, how do you see AI tools complementing your decision-making process without undermining the relationship-driven nature of investing?
AI tools can enhance the decision-making process of early-stage investors by increasing efficiency through automated market screening, quick feedback on investment opportunities, and automated reporting. However, the use of AI may impact the personal touch and relationship-building that are crucial in the investment process, as well as the human intuition that plays a vital role in assessing founders and their visions. Additionally, the effectiveness of AI is limited by the availability and quality of data. Implementing and maintaining AI tools can also be costly and complex. Despite these drawbacks, the potential benefits of AI in enhancing efficiency and data analysis make it a valuable complement to the decision-making process, provided that its limitations are carefully managed.

So, what's next?
While AI is still in its early days in the venture world, it's clear that it's here to stay. Investors are beginning to use it to streamline operations, analyze data more efficiently, and identify opportunities that may have otherwise been overlooked.

Entrepreneurs, fear not—at least for now. While algorithms can process scale, they can't yet read between the lines. The final call still comes down to a human reading a human. The future of venture capital will continue to depend on trusted relationships—areas where technology still has a long way to go.

As AI becomes part of the venture capital process, the challenge will be to use it for efficiency without losing the human judgment that matters most. For now, AI serves as a tool, not a substitute, for experienced investors. As Andreas and Michelle point out, while AI can highlight patterns and organize information, the final decisions come down to people. When it comes to funding the next generation of unicorns, relationships remain key.