Source: Lenny’s Newsletter by Lenny Rachitsky. Read the original article
TL;DR Summary of How CJ Hess Builds Custom AI Dev Tools and Uses Model-vs-Model Code Reviews
CJ Hess demonstrates how he leverages AI to automate over 90% of front-end coding by managing AI teams rather than writing code manually. He combines Claude Code’s intent understanding with GPT-5.2 Codex’s review capabilities for superior code quality through a model-vs-model workflow. Additionally, CJ uses custom visual planning tools like Flowy to reduce cognitive load and improve development efficiency, highlighting the importance of adapting AI tools to human preferences.
Optimixed’s Overview: Leveraging Dual AI Models and Visual Planning for Advanced Front-End Development
Introduction to CJ Hess’s AI-Powered Development Workflow
Engineer CJ Hess from Tenex shares a practical approach to integrating AI into software development, focusing on front-end coding. By building custom workflows and tools, he has successfully delegated over 90% of his coding tasks to AI, significantly boosting productivity and innovation.
Key Components of CJ’s AI Workflow
- Dual-Model Strategy for Code Generation and Review: CJ pairs Claude Code, which excels at understanding coding intent, with GPT-5.2 Codex, which specializes in code review and quality assurance. This collaboration ensures faster and more reliable code output than relying on a single AI model.
- Custom Skills to Enhance AI Capabilities: He develops specific “skills” that teach Claude how to generate accurate JSON for tools like Flowy, enabling the AI to produce detailed flowcharts and UI mockups tailored to his projects.
- Visual Planning to Reduce Cognitive Load: Instead of relying solely on ASCII diagrams, CJ uses Flowy’s interactive visualizations. This method aligns better with human cognition, making it easier to review and approve design plans.
- Managing AI with Human Oversight: CJ employs an alias with elevated permissions to allow AI systems to work more autonomously but maintains safeguards through Git, balancing risk with efficiency.
Benefits and Impact
This approach demonstrates the power of combining AI tools thoughtfully to maximize their strengths. CJ’s workflow enables developers to focus on higher-level problem solving while AI handles routine coding tasks, illustrating a scalable and practical path toward AI-assisted software development.