Google’s former Chief Scientist, Jeff Dean, was interviewed by Diana Hu of Y Combinator, where he explained that the AI model used is increasingly less important than how it is used within a larger system of tools, retrieval, and AI agents. Dean agreed with Hu that progress is shifting toward “context engineering,” which involves everything around the model, such as retrieval, tools, memory, and agent tools, rather than just model size, parameters, or data. He stated that the model is only one piece of building an overall system that can solve interesting problems, and that it involves a model that knows how to use various tools, retrieve relevant information, and put information into the model’s context. Dean also emphasized the growing importance of orchestrating multi-agent systems, which means coordinating AI agents for how they use tools and retrieve relevant information to solve complex problems. He contrasted an AI model with all its training data against an AI looking at a collection of information directly relevant to its task, leading to the point that a model is better able to do a job when it has the right level of orchestration, and that this is where things are headed.