Why this Claude Code engineer uses HTML files as AI specs | Thariq Shihipar (Anthropic)

2026-05-18 · Source: How I AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Emerging Technologies & Innovation · Depth: Intermediate, extended

Summary

Claraveo and Thariq, a product leader and a Claude Code team member, discuss the emerging trend of using HTML instead of Markdown for interacting with AI agents, particularly Claude. They highlight that while Markdown was popular, the increasing length and complexity of agent-generated plans (often 50-line code plans for Opus 4, extending to hour-long runs for Opus 4.5 and 4.7) make them difficult for humans to read and edit. HTML offers a richer, more engaging communication medium, allowing for visual mockups, interactive elements, and custom UIs that improve human engagement and overall output quality. The discussion emphasizes the continued importance of detailed specifications and planning, framing product managers as "compute allocators" who must decide how to best utilize expensive AI compute resources. They demonstrate how Claude Code can generate HTML-based brainstorming ideas, detailed implementation plans with mockups and code excerpts, and even throwaway custom UIs for editing specific plan sections.

Key takeaway

For AI Architects designing agentic workflows, recognize that HTML-based interactions with models like Claude Code can significantly enhance human oversight and collaboration. You should explore generating dynamic, visual plans and custom UIs in HTML to improve engagement with complex outputs, ensuring better quality and more efficient allocation of compute resources. This approach allows for more intuitive editing and a clearer understanding of agent actions, moving beyond static Markdown limitations.

Key insights

HTML is replacing Markdown as the preferred medium for human-AI interaction due to its richer, more engaging visual capabilities.

Principles

• Human engagement improves AI output quality.
• Product managers are now compute allocators.
• Specs and plans remain critical for AI projects.

Method

Use Claude Code to generate HTML-based brainstorming ideas, detailed implementation plans, and custom UIs for interactive editing, fostering better human-agent collaboration and higher quality outputs.

In practice

• Prompt Claude to "brainstorm ideas in HTML."
• Create custom UIs for specific editing tasks.
• Encode design systems in HTML for easy reference.

Topics

• HTML AI Specifications
• Claude Code
• Compute Allocation
• AI Agent Interaction
• Custom UI Generation

Best for: AI Architect, AI Engineer, AI Product Manager, Director of AI/ML

Related on AIssential

• I hated making this video... — Claude Code innovates AI agent interaction and capability through script execution, modular configuration, and parallel processing.
• According to Anthropic’s Engineer, HTML is the New Markdown for AI Agents — HTML transforms static AI outputs into interactive, reusable interfaces, enabling human action.
• Anthropic’s Claude AI can respond with charts, diagrams, and other visuals now - The Verge — Claude now generates dynamic, in-line visualizations to enhance conversational understanding and interaction.
• I Built My Friend a Free Personalised Fitness Plan Online — Without Writing a Single Line of Code — AI can rapidly generate and deploy functional web applications from detailed prompts.
• Anthropic’s Code with Claude showed off coding’s future—whether you like it or not — LLMs are rapidly automating software development, shifting human roles from writing to oversight, despite emerging concerns.
• Dive into Claude Code: The Design Space of Today's and Future AI Agent Systems — AI agent architectures are shaped by human values and deployment context, even with similar core functionalities.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by How I AI.