Investing in the Shadow of AGI: Merge the old and the new

VCs are famous for asking startups, “What is stopping Apple, Meta, and Alphabet from doing this?”

The new question VCs are thinking about, if not asking explicitly, is “What is stopping general AI from just doing this?”

When we invest in seed-stage companies, we know that these startups may take a decade or more to reach M&A maturity or the public markets. Predicting what will happen next quarter, never mind five years from now, feels like a fool’s errand. The speed of disruption has never been faster.

• ChatGPT has transformed how we think about text over the last two years.
• Midjourney has helped people imagine an entirely new way to create images.
• Suno (an FC portfolio company) launched less than six months ago before becoming one of the top five AI appsin the world.

This pace of change is a problem for software startups currently raising seed rounds. If you’re not using AI, and using it in a powerful way, there’s a solid risk that AI will fundamentally disrupt what you’re building. Even if you are using AI, the general model may come for your lunch too, e.g. Jasper.

Herein lies the conundrum – founders and funders can’t focus exclusively on AI, nor are “traditional” SAAS companies en vogue. A pitch for a new software company has to be more than “with AI, we can do X.” Founders need ingredients beyond AI, because the models are rapidly commoditizing and readily available. Some of these additional attributes include:

• Deep domain expertise for specific verticals
• An interface layer tailored to specific users needs (beyond the chat client)
• Proprietary data sets
• Hardware
• Exclusive alliances and partnerships
• Specialized knowhow
• Patents

We hope entrepreneurs will continue to pursue ideas outside of AI. During the mobile revolution, the last big tech upheaval, Honey built a multi-billion dollar business building browser plug-ins. Figma was primarily a web app. Shield AI was building hardware and AI (before it was cool). Plenty of $1B+ startups will be built that don’t showcase AI or where its only a component of a larger feature set.

While AI dominates the conversation, the models themselves may face their own demise in this rapid cycle of innovation. The question is can you grow (or sell) before the next tool renders the last obsolete. Or better yet, build a sustainable model that’s more than just a wrapper around AI.