The Death of the Blank Canvas
How AI agents are turning designers into directors
The traditional design process is a slow climb from a blank Figma file to a polished prototype. It requires intent, planning, and a significant amount of manual labour. But for John Bai and Peng Zheng, designers at SpaceXAI, that era is ending. They are not merely using AI to suggest colours or layouts; they are building entire ecosystems of autonomous agents that handle the heavy lifting of production. This shift changes the designer's role from a craftsman of individual elements to a conductor of digital systems. The friction between an idea and its execution is evaporating, replaced by a loop of rapid experimentation and automated refinement.
The Self-Updating Life
Peng Zheng has abandoned the static portfolio. Instead of building a website that sits idle between job applications, he has engineered a Grok Bot to serve as his entire backend pipeline. The workflow is deceptively simple: Peng sends a photo or a location name to the bot, and the machine takes over. It identifies the coordinates via Google Places, generates custom 3D miniature artwork of the location, and publishes the update to his site. There is no CMS to log into and no Figma file to update manually. The site lives and breathes alongside his actual movements, turning his life into a continuous, automated stream of content.
The cost of making things has fallen so dramatically that unusual experiments are finally worth the attempt.
This isn't just about convenience; it is about the removal of the 'planning tax'. In traditional product development, you often need a roadmap and a specification before you can even begin to see if an idea works. Peng's approach ignores this. He starts with a vague description, reacts to what the bot produces, and follows the most interesting paths. Because the bot handles the architecture, the designer is free to wander. This allows for a type of emergent design where the final product is not a direct result of a pre-conceived plan, but a series of successful reactions to unexpected outputs.
Design from the Gym
John Bai has taken this even further by decoupling design from the desk. Through a 'Figma Bro' bot, he uses voice memos to direct complex production tasks while he is away from his computer. If a colleague needs assets while he is working out, he can simply record a messy, natural-language instruction. The bot interprets the intent, accesses Figma through an MCP connection, and executes the work—adjusting sizing, colours, and artboard structures according to pre-configured preferences. The designer is no longer tethered to the mouse and keyboard; they are managing a workforce of digital specialists.
- Narrowly defined agents for specific tasks like email or 3D rendering
- Voice-to-action workflows that bypass manual UI interaction
- Rapid prototyping loops that move from 'shower thought' to working model in minutes
- Automated backend pipelines that eliminate the need for manual content management
As these tools mature, the barrier to entry for complex creative skills—like 3D modelling or motion design—will continue to drop. The value will shift away from the ability to operate software and toward the ability to direct intent. The question for the next generation of designers is not 'can you draw this?' but 'can you describe this well enough for an agent to build it?'
The designer's future lies in directing autonomous agents rather than manually manipulating pixels.