Only about 1% of the tokens Thariq generates go into production code

2026-05-21 · Source: How I AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, quick

Summary

Thariq, an individual generating tokens, reports that only approximately 1% of the tokens they produce are integrated into production code. The vast majority are utilized for exploratory purposes, such as developing dashboards and custom interfaces, to refine ideas and plans. This contrasts sharply with traditional software development practices, where significant effort was expended on maintaining a single "source of truth" for specifications and PRDs across consistent formats and templates. The author posits that as the cost of content creation, consumption, and discovery approaches zero, the rigid adherence to such arbitrary rules diminishes. This shift allows for a greater focus on the quality and substance of a plan, fostering an environment for "just-in-time documentation" and "throw away software" that is inexpensive and disposable. The author also highlights the importance of interacting with aesthetically pleasing outputs, aiming for a more integrated and collaborative development process, exemplified by working with tools like Claude.

Key takeaway

For AI Engineers or product managers developing new features, you should re-evaluate traditional documentation overhead. Embrace AI-assisted content generation to rapidly prototype ideas and interfaces, allowing you to prioritize the quality of your plan over rigid format adherence. Focus on "just-in-time documentation" and be willing to create "throw away software" for exploration, significantly reducing development friction and accelerating iteration cycles.

Key insights

AI-driven content generation reduces documentation costs, enabling focus on idea quality and flexible, just-in-time practices.

Principles

• Reduced content cost allows format flexibility.
• Prioritize plan content over rigid templates.
• Embrace disposable, "throw away" software.

Method

Rapid prototyping and exploration using AI to generate numerous tokens for interfaces and dashboards, followed by just-in-time documentation.

In practice

• Use AI for rapid dashboard prototyping.
• Generate diverse content for exploration.
• Adopt just-in-time documentation.

Topics

• AI-assisted Development
• Token Generation
• Just-in-Time Documentation
• Software Prototyping
• Development Workflow
• Claude AI

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

Related on AIssential

• 😺 Amazon spent $200B and broke its own website — AI-generated code can introduce significant instability and operational risks if not managed with rigorous human oversight.
• Use AI for writing without the cleanup tax — Proactive prompt engineering can eliminate common AI writing flaws and reduce post-generation editing significantly.
• Why AI types one word at a time — Chatbots generate responses token by token, predicting the next most probable element iteratively.
• ‘Tokenmaxxing’ is making developers less productive than they think — Increased AI code generation correlates with disproportionately higher code churn and revision rates, undermining productivity claims.
• I Let ChatGPT Write Production Code for 30 Days. It Cost Me $127,000. — Blindly trusting AI for production code without human oversight leads to significant cost overruns and system failures.
• How to Avoid AI Code Slop — AI-accelerated development necessitates shifting code quality gates upstream to intent-driven verification, not post-code review.

Open in AIssential →

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