Voice-Driven Development: Building a Blog Feature While Cooking
Voice AI
Coding Agents
LLMs
Developer Workflow
Multitasking
Codex
This summary and analysis were generated by AI from the original article at Simon Willison and may contain errors (how Viqus works). Read the source for full details.
7
Productivity Leap in Agentic Coding
Media Hype
6/10
Real Impact
7/10
What is the Viqus Verdict?
We evaluate each news story based on its real impact versus its media hype to offer a clear and objective perspective.
AI Analysis:
The hype is high due to the narrative, but the real impact lies in the demonstrable capability to manage complex state changes (like database migrations) via voice, marking a genuine workflow shift.
Article Summary
Simon Willison detailed the process of developing a new newsletter index feature for his personal blog, accomplishing the task almost entirely through voice conversation with the ChatGPT desktop app's Codex mode. He guided the AI through defining the data model, creating migrations, and writing necessary view code, all while cooking dinner. The AI successfully handled complex logic, such as differentiating between searchable and non-searchable content sources (Substack vs. internal newsletters). While the initial draft required manual intervention—specifically swapping out a subprocess Git call for a more robust API-based import—the core functionality, including integration with site search and multiple data sources, was achieved through voice prompting. The author concluded that the visual preview and multi-tasking capability make this interaction significantly more powerful than current phone-based use cases.Key Points
- The author built a functional, multi-source newsletter index page using only voice commands directed at an AI coding agent.
- The process demonstrated the agent's ability to handle complex backend tasks, including database modeling, API integration, and content source differentiation.
- The key improvement over previous voice interactions is the combination of voice input with a visual local development preview, enabling iterative refinement.