November 29, 2022 | 5 minutes read

You know the “Why.” Here’s the “How”

Building a Future-Proof Generative AI Startup

Nir Sabato
Investor at Entrée Capital

Over the last few months, we’ve seen hundreds of articles about generative AI. Almost all of the content has so far focused on the “why part of generative AI. Why it matters, why now, why it will grow. So, we decided to focus on something a bit different: the “how.” 

As a founder, how can you be sure that your generative-AI offering will be “future-proof”? How can you build a relevant business? How can you prove to investors that your idea is investment-worthy? Here are some of our thoughts:

🎯 Get the problem and opportunity right

Don’t think abstractly: The benefits of generative AI are widespread, but it won’t do you any good to think abstractly about the problem or the opportunity. Start with a deep understanding of the industry. Stay away from building something that will only solve “trivial” problems (like content generation for content creators) or something that’ll simply be another “fun to play with” tool rather than an actual product that solves current needs.

Find real-world problems: Invest the time to identify underserved client segments and real-world issues that can be disrupted by a killer product and win. Also, think of niche challenges and build something that goes above and beyond the competition by utilizing AI capabilities. Remember that the “AI” part of the app is just one of the building blocks that needs to be included in the product innovation. 

When choosing this “space,” take the traditional approach of identifying a big market with low NPS and a high level of fragmentation, and be better in that market because of AI. In short: start with the problem, not with the technology. 

🛠 Build a relevant product

Focus on application, not infrastructure: The infrastructure layer of generative AI is growing rapidly and is fueled by massive funding from big tech and deep pockets investors (think, Midjourney, Stable Diffusion, GPT-3, etc.) These AI “infrastructure-level” models and capabilities will continue to change rapidly over the next few years. So, unless you’re a complete expert in the space and can raise tons of cash, don’t spend your efforts building infrastructure that may not be relevant (don’t bring a knife to a gunfight). 

Instead, focus on the application layer, where you can take the models and apply them to a product that can help humans solve a real existing problem in an innovative way. Ask yourself: 

  • Who is the end user and what problems are they encountering? 
  • How could a “Generative AI” layer create unique value for this end user?
  • What features aside from AI are necessary to create a differentiated product?
  • What type of interaction should there be between the user and the product? Can the user fine-tune, create automations and workflows, and teach and train the model?  
  • How naturally and seamlessly can an existing (or new) AI model be part of the product? 

🔮 Be future-proof

Don’t rely solely on AI: As we said earlier, the “winners” in the space won’t be the players with the “strongest AI model” but rather the ones that leverage the AI capabilities and integrate them into a “killer app” to beat out the incumbents or solve a billion-dollar problem. What may seem like “magic” today will be a common standard 2-3 years from now. Any business that relies on the generative model as the only means to create value will be less defensible in the long run. 

Don’t be a single-algorithm company: Don’t build on just one model. Use a collection of technologies and build in a way that you can plug in/out components or use cases. Start with a focus on a specific use case or capability. Iterate to build a platform over time that will know how to introduce and deal with new use cases too. 

Think long-term: Balance out short-term revenues against long-term vision. It may be highly tempting to see today’s technology and think of immediate or short-term use cases, but by focusing on the short term and neglecting the long term, you may find yourself trying to push an outdated solution that may soon be commoditized. Take into account that what might look like a blue ocean now might turn red very soon.

We at Entrée Capital have already invested in startups utilizing generative AI like Bria, Darrow, Swapp, and more. We are actively looking to back more “future-proof” startups with teams that understand how generative models can be a once-in-a-decade disruptive power across all businesses. Show us the big picture and the big plans ahead, and we can help you get there. 

(And no, this blog was not written with AI, but a big thanks to Yair Adato of Bria for your input). 

🔥 Want to hear from founders, CEOs, and academics in the generative AI space? Join us at our upcoming event on December 21st at Facebook Tel Aviv: Generative AI | Building to Last