July 2023
Alline Akintore
We've been refining our thinking on investing in Generative AI. We often collaborate with other investors to share knowledge and insight. We held an AI workshop in early July which provided a valuable opportunity for us, along with venture partners and portfolio company leaders, to further our learnings and considerations for venture investing in AI companies.
Special thanks to our panelists and speakers, Geoff Harris, Flying Fish Venture Partners; TA McCann, PSL Ventures; and Tim Porter, Madrona Venture Group whose insights contributed to the takeaways featured here.
Takeaways for founding and investing in early-stage AI startups
1. When evaluating AI startups, what to look out for:
Is the startup generative assistive or generative native? How core is AI to the company’s sustainable differentiation and competitive advantage?
Are the founders using the technology in a truly differentiated way? Is there an opportunity to create IP?
Can a community grow around the software?
Does the company have access to proprietary data? Creating or capturing their own unique data?
Is the team expert with both AI and with their market? Ideally, they have expertise in both.
2. Find white space opportunities, such as the opportunity to create foundation models for specific verticals. Startup opportunities may exist between dominant platforms, such as workflow solutions that might involve, for example, both Google and Salesforce that neither company would likely offer on their own.
3. In the future, an opportunity may lie with “Reinforcement Learning,” where AI agents can learn autonomously, without any existing training data. One would be able to drop an AI agent into the world – with no prior training – and it would learn to be productive/valuable on its own.
4. A common question is whether all the value created from this generation of AI will accrue to incumbents. LLMs today are like cloud infrastructure 10 years ago. The largest players with current data advantages will accrue huge financial benefits, but they will also offer and monetize their AI services for the benefit of thousands of startups, much like the cloud era.
5. Pay attention to the cost of delivering an AI-based service. With fully loaded costs, the business may not pencil. Cost to deliver (CTD) could impact the economics of the business.
AI technology and broader market takeaways
1. There is no AI “moat.” There will be thousands of AI models and versions of those models. It could make sense to use existing infrastructure like Azure AI Services or other cloud AI platforms, which are good enough. No need to reinvent the wheel.
2. AI will enable startups to better target and serve very specific customer profiles. They will also qualify customer leads faster and better.
3. Today the investing market for deep AI startups is frothy. Is that rational? Maybe, if the company has potential to achieve an early mover and market advantage.
If you are an early-stage startup founder or executive building an AI company in Oregon or Southern Washington, please reach out to us. We’d love to meet you and be helpful.