The author, while admitting limited experience with AI, reflects on the design of AI tools and their shortcomings. They outline a typical interaction flow with AI: asking questions, waiting for responses, and analyzing the answers. This process can become tedious and costly, especially when continuous queries are involved. For developers, the interaction is slightly different; they integrate AI into their workflow but still face latency issues that can hamper productivity. Three key principles for effective tool design emerge from the author's perspective: low latency, direct manipulation, and affordability. Low latency minimizes the time between action and feedback, which is essential for maintaining focus and efficiency. Direct manipulation allows users to see results in real time, reducing the lag between input and output. Affordability ensures long-term access to tools without the anxiety of subscriptions or software obsolescence. The author argues that current AI tools often fail to meet these criteria, leading to frustration and inefficiency. As a counterpoint, the author highlights a cutting-edge AI model that allows for direct interaction with video content. This model boasts lower latency and is open source, ensuring its usability in the future. The author expresses optimism for local models over current hosted solutions, emphasizing their potential for better user experience. They also note that many existing systems introduce complexity to compensate for slow response times, which increases mental overhead for users. By focusing on direct manipulation and reducing latency, the author believes AI tools can become more effective and user-friendly, allowing for a more integrated workflow.