The creator compares two AI models, Fable 5.1 and GPT-6 Astra, in building native Mac clipboard managers, highlighting how Fable offers thoughtful, detailed designs while Astra excels in rapid iteration and usability improvements. They emphasize the value of using multiple AI models to leverage their unique strengths, resulting in more diverse, refined, and practical software solutions.
In this video, the creator compares two AI models, Fable 5.1 (Claude) and GPT-6 Astra, by giving them the same prompt to build a native Mac clipboard manager aimed at solving the common problem of “copy-paste hell.” Both models produced functional but distinctly different apps: Fable created “Ledge,” a narrow list sliding in from the right, while Astra developed “Shelf,” a wider bar across the bottom with larger cards and a preview pane. Despite the same brief and time constraints, the two apps showcased unique design philosophies, highlighting how AI can offer diverse solutions to the same problem.
The creator emphasizes the importance of the second round of iteration after receiving the initial build. Astra’s faster response time and lower token usage allowed for more rapid revisions and improvements, such as repositioning the panel, adjusting hotkeys, and adding confirmation messages for successful copy actions. These refinements made Astra’s app more usable and intuitive, demonstrating how speed and iteration capability can significantly impact the quality and practicality of AI-generated software.
A key takeaway is that the differences between the models are not about intelligence but perspective and approach. Fable asked clarifying questions and provided a thoughtful, detailed design, while Astra focused on speed and usability. The creator found value in using both models to explore the solution space, discovering preferences and refining requirements through hands-on experience rather than relying solely on initial prompts. This approach allowed for a more nuanced understanding of what was needed in the final product.
The video also discusses the broader implications of using multiple AI models for different tasks. For example, Fable excels in deep problem thinking and design, making it suitable for initial brainstorming and complex planning, while Astra shines in rapid iteration, coding, and integration with tools like MacOS. The creator advocates for leveraging the strengths of each model depending on the task, rather than seeking a single “best” model, and highlights the potential of multi-agent systems to coordinate these strengths for even more effective outcomes.
Finally, the creator reflects on the current state of AI-assisted software development, expressing excitement about how quickly and easily useful apps can now be built with minimal prompts and iterations. They encourage viewers to try both apps, “Shelf” and “Ledge,” and to consider using multiple AI models to gain diverse perspectives and better results. The video concludes with a promise to continue exploring and comparing AI models, sharing insights to help others harness these powerful tools effectively.