The Missing Layer: Design Taste in AI Agents — Hassan El Mghari, Together AI
Source
youtube.com
Author
AI Engineer
Date
Context
Hallmark turns design preferences into reusable constraints and reference themes for coding agents. In the publisher's description of the talk, Hassan El Mghari, who leads developer experience at Together AI, explains the approach through recognizable tells of generated apps such as purple gradient backgrounds and all-caps pills.
The described workflow combines naming those patterns as exclusions with a library of themes and a vault of screenshots the user admires. Accumulated preferences are written into a agent skillA reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.Full definition → file or AGENTS.md for later work. The description presents what the AI agentAn 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.Full definition → returns as a base to iterate on, not a finished interface.
Terms in this piece · Glossary
agent skill — A reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.
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.