View all videos

The future of software is AI, the future of AI is software

Eiso Kant, Co-Founder & CTO,, PoolsideJason Warner, Co-Founder & CEO, Poolside

On Focusing on Software Development:

The speakers highlighted their departure from general-purpose AI models like GPT-4, emphasizing a concentrated approach on the singular capability of software development. This strategic choice involves honing in on the unique challenges and opportunities presented by the world of coding. Quote: "We're taking a different path than some of the general-purpose approaches that you might have seen in models like GPT-4... really focusing on a single capability, the capability of software development." On the Unique Role of Neural Networks in Code Generation: The speakers articulated the distinctive role of neural networks in generating code, pointing out that code serves as an "oracle of truth." Unlike other tasks where human feedback is crucial, the executable nature of code allows for direct validation, making it a valuable resource for training AI models. Quote: "One of the only things that neural networks can generate that we don't need human feedback on, but that we can actually run, introspect, and execute is code, which means that we have an actual oracle of truth."

On the Future Vision of AI-Led Software Development:

The speakers painted a vivid picture of the future, where AI leads the software development process, and humans assist in a collaborative, conversational manner. This shift envisions a world where individuals can convey high-level software ideas to an AI system, initiating a dynamic exchange that results in the creation of fully functional software. Quote: "So for us, it's a perfect bubble in which we can test things... because the vast majority of code is just that. It's just a language that we've created to do structured reasoning to deal with data, to deal with effects on environments."