Imagination Engineering: Building Software on Demand
AI Engineergo watch the original →
the gist
Eve Bouffard demonstrates using LLMs to rapidly prototype custom tools and personal dashboards, arguing that the new bottleneck for developers is creative ideation rather than technical execution.
Thinking in Public as a Development Workflow
Eve Bouffard advocates for "thinking in public" by dumping raw streams of consciousness into a dedicated Slack channel and using LLMs to aggregate these thoughts into functional, interactive web applications. By treating personal thoughts as a data source, she uses models like Claude 3.5 Sonnet (referred to as Opus 4.8 in the talk) to build personal websites that automatically reflect her current projects, reading lists, and design preferences. This process includes generating a "voice MD" file that captures her unique tone and writing style, which the agent then uses to maintain consistency across new components and design updates.
Rapid Software Prototyping
Bouffard demonstrates that modern LLMs enable "software on demand," where custom tools are built in minutes rather than days. Her "Shape of Minds" project, a visualizer mapping the commonalities of historical geniuses, was built in a single morning. The tool uses a responsive UI to display data points like habits, work ethic, and quirks of historical figures. She also builds micro-tools for daily productivity, such as a custom emoji picker for Slack that bypasses the limitations of the standard iOS interface. These projects rely on agents to handle the heavy lifting of code generation, allowing the user to focus on defining the goal and the desired design language.
AI as a Cognitive Accelerator
Beyond building tools, Bouffard positions AI as a "rocket ship for the mind" that facilitates rapid learning. By requesting summarized reports on complex topics—such as Renaissance patronage or design principles—she creates a personalized curriculum that is more efficient than traditional research. She emphasizes that the current era of AI capability mirrors the Library of Alexandria, where the primary challenge for humanity is the consolidation and creation of knowledge. Her workflow involves spinning up agents to handle specific tasks on the fly, effectively treating the LLM as an extension of her own cognitive process.