The true power of the AI application layer lies in orchestrating multiple, specialized foundation models. Users want a single interface (like Cursor for coding) that intelligently routes tasks to the best model (e.g., Gemini for front-end, Codex for back-end), creating value through aggregation and workflow integration.

Stripe avoids costly system rebuilds by treating its new payments foundation model as a modular component. Its powerful embeddings are simply added as new features to many existing ML classifiers, instantly boosting their performance with minimal engineering effort.

Instead of being a weakness, Cursor's reliance on multiple foundation models is a key strength. With 50% of developers switching model families daily, this approach allows Cursor to benefit from every improvement in any underlying model. This creates a compounding product flywheel, making the application layer an index of the entire AI ecosystem's progress.

Low-Rank Adaptation (LoRa) allows a single base AI model to be efficiently fine-tuned into multiple, distinct specialist models. This is a powerful strategy for companies needing varied editing capabilities, such as for different client aesthetics, without the high cost of training and maintaining separate large models.

Rather than committing to a single LLM provider like OpenAI or Gemini, Hux uses multiple commercial models. They've found that different models excel at different tasks within their app. This multi-model strategy allows them to optimize for quality and latency on a per-workflow basis, avoiding a one-size-fits-all compromise.

By making different foundation models (like Gemini and Claude) collaborate, developers can achieve superior outcomes. One model's unique knowledge, such as using a free RSS feed instead of costly APIs, can create vastly more efficient and creative solutions than a single model could alone.

The initial AI rush for every company to build proprietary models is over. The new winning strategy, seen with firms like Adobe, is to leverage existing product distribution by integrating multiple best-in-class third-party models, enabling faster and more powerful user experiences.

The belief that a single, god-level foundation model would dominate has proven false. Horowitz points to successful AI applications like Cursor, which uses 13 different models. This shows that value lies in the complex orchestration and design at the application layer, not just in having the largest single model.

Criteo's models moved from using manually crafted, extremely high-dimensional sparse vectors (e.g., 2^12 features) with linear models to dense vectors (a few hundred features) automatically computed by deep learning algorithms. This shift eliminated manual feature engineering and improved model adaptability.