The author explores the concept of "context plumbing" in AI systems, focusing on how understanding user intent and context enhances the effectiveness of AI interfaces. Intent refers to the user's goals, which can be expressed explicitly or implicitly. AI has the ability to interpret these intents in a human-like manner, allowing for smoother interactions. The shift from desktop to smartphone illustrates this; on phones, users can interact directly, reducing the cognitive load of navigating through menus or interfaces. Companies aim to harness this by developing devices like AI-enabled glasses that minimize user effort and friction. A key factor in understanding intent is context. AI systems perform better when they have access to dynamic context, which can include background knowledge, user history, and the current environment. The author introduces the idea of context engineering, where systems are designed to deliver relevant information in the right format for the AI to act effectively. He emphasizes that the AI needs to be close to where user intent arises and should be continuously updated with the latest context to avoid slow responses. This leads to the analogy of plumbing, where the challenge lies in efficiently moving context data from various sources to where it is needed. The article reflects on the evolution of technical architecture from Web 2.0 CRUD applications to AI systems that require a more intuitive understanding of context. The author is currently building a platform on Cloudflare that effectively manages context flow among various AI agents without complexity. While details about the platform remain under wraps, the author notes that it has taken two years of development to reach this functional stage. The focus is on ensuring seamless integration and user expectation alignment in the context flow of AI interactions.