Vibeleaderboard
← All Intel
Intel / video

How we Claude Code

Source
youtube.com
Author
Claude
Date

Context

Anthropic's How we Claude Code workshop makes one concrete proposal: put the plan an will work from in an HTML file rather than a long markdown file. The presenter, an architect on Anthropic's Applied AI team, bases it on a talk by Thariq of the Claude Code team, published as the blog post The Unreasonable Effectiveness of HTML. Markdown specs beyond roughly 200 lines, the presenter says, are unlikely to be read, while a web page can carry mockups and layouts a person can react to. The demo asks Claude for four HTML design directions for a bill-splitting app so they can be compared before anything is built.

The stated reason is cost control: the longer an agent runs, the more a misunderstood spec wastes, so more human checking moves to the start. The presenter says HTML is not necessarily more expensive overall because a rich spec cuts iteration, though generating it once costs more; no measurements were given.

Terms in this piece · Glossary
  • AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Read the source www.youtube.com
More from Claude
Recommended reads
Comments

Checking sign-in…

Loading comments…