Developer documentation has traditionally been designed for people navigating pages in a browser. AI agents and code assistants use it differently: they need to locate, interpret, and turn documentation into working implementation context while a developer is actively coding.
This session shares what server-log analysis of more than 10 million requests to Adyen’s documentation revealed about agent traffic, including how frequently user-directed assistants retrieve context. It then explains how the documentation layer was adapted for this workflow with structured LLM discovery, raw Markdown URLs, and a downloadable documentation mirror.
What the session covers
- Why human-first HTML and site hierarchies create unnecessary traversal for agents
- How
/llms.txtsupports structured content discovery - How raw
.mdURLs reduce parsing and reconstruction overhead - How a complete Markdown mirror can provide portable documentation context
- How agent-ready delivery can reduce rate-limit pressure and improve generated integration code
Session details
This intermediate Agentic Engineering breakout takes place in room 210 CG on Friday, October 23, 2026, from 5:10 PM to 5:35 PM PDT.