Ryo Lu, Head of Design at Cursor, demonstrates building a functional app (a calculator) within his personal 'RyoOS' project entirely through AI prompts in Cursor, bypassing traditional design tools like Figma.
The discussion highlights a shift in software development workflows, where AI enables designers to prototype and build directly in code, significantly reducing the gap and friction between design and engineering.
Cursor's internal product development process is highly agile, eschewing formal roadmaps and dedicated PMs in favor of blurred roles, rapid iteration, and a tight feedback loop with a 'nightly users' group.
The evolution of AI tools like Cursor is enabling higher levels of abstraction in coding, such as 'Plan Mode,' allowing users to focus on architecture and intent rather than line-by-line implementation.
12 quotes
Concerns Raised
AI-generated code still requires human oversight and debugging, as demonstrated by the agent running into an error during the live demo.
The rapid advancement of models (e.g., Gemini recreating a project in one shot) could quickly make existing tools and workflows obsolete.
The transition from traditional design tools to a code-first AI workflow requires a significant skill shift for many designers.
Opportunities Identified
Massive productivity gains by collapsing the design-to-development cycle.
Empowering designers and other non-engineers to build and ship functional software independently.
Creation of more dynamic, integrated, and agile product teams by blurring traditional role definitions.
Ability to rapidly prototype and iterate on complex applications with high-level, natural language prompts.