Stop building LLM wrappers and aggregators, says Google VP (techcrunch.com)

Darren Mowry, Google’s VP overseeing its global startup initiatives, has issued a warning to emerging companies in the AI space, particularly those building LLM (Large Language Model) wrappers and aggregators. He emphasized that startups relying heavily on existing models, like GPT or Gemini, without offering significant differentiation or unique intellectual property are facing a challenging market. Mowry highlighted that the landscape is shifting towards sustainable product value, reflecting an industry trend where consumers prioritize functionality and innovation over basic access to LLMs. Startups must now focus on creating deep, unique offerings rather than merely layering a UI on existing technology. Mowry's insights draw parallels to the early cloud computing era when startups that simply resold AWS infrastructure struggled to survive once Amazon expanded their services. Similarly, AI aggregators, which provide access to multiple models through a single interface, are under pressure as model providers develop enterprise-level features directly. Mowry advocates for a shift toward “vibe coding” and developer platforms that empower users and create bespoke solutions, alongside growth opportunities in direct-to-consumer technologies and sectors like biotech and climate tech that leverage vast data resources for impactful results.