Mehdi Medjaoui (Apidays): All your docs are belong to us: from UX to DX to AX

This talk was presented at the AI The Docs 2025 online conference. We are thrilled to share the recording and the summary with you. 

Visit the talk summary page to see all of the presentations from the conference.

presentation

Mehdi Medjaoui (Founder & Chairman of Apidays conference)

Summary

How can technical writers prepare documentation for an AI-first world where both humans and agents consume content? 

Keynote speaker Mehdi Medjaoui (Founder of Apidays conferences series) traces the evolution of technical writing from Scribes (documenting intent), to Industrial Age writers (documenting exact instructions), to Computer Age writers (maintaining a single source of truth). 

As Mehdi Medjaoui sees, now, documentation enters a new phase: serving not just humans, but AI agents. 

• The opportunity:
  • AI can generate first drafts, overcoming “blank-page trauma”.
  • Docs become runtime interpreters for LLMs, enabling embedded search, test generation, and chatbot interfaces.
• The limitations:
  • AI lacks empathy, pedagogy, and audience awareness.
  • It cannot decide what not to document (an essential part of design intent).
  • Generated content risks being generic or misaligned with reader needs.
• Practical recommendations:
  • Audit docs for LLM readiness (structured, modular, machine-parsable).
  • Provide prompt libraries alongside code samples.
  • Experiment with embedded search and chatbot interfaces.
  • Measure impact by successful queries, not clicks.
• The mindset shift:
  • From Docs as Code → Docs as Context: AI can and will read everything, so writers must ensure content is semantically structured and in the right context.
  • From User Experience (UX) → Developer Experience (DX) → Agent Experience (AX): docs must now support agents as well as people. 

As Mehdi highlights, 

“The future is not ‘Docs as Code’, it’s ‘Docs as context’’. 

Principles of Agent Experience (AX): 

• Human-centricity: agents must serve real human intent.
• Contextual alignment: ensuring data and use cases are clear.
• Accessibility for agents: structured pathways, identity, and authorization.
• Dedicated interactivity patterns: separate paths for agents vs. humans. 

Future role of technical writers: 

• Structure everything semantically (one concept per block).
• Master API description standards (OpenAPI, AsyncAPI, JSON Schema).
• Create dual experiences: for developers (DX) and for agents (AX).
• Provide prompts and examples as training data.
• Monitor agent performance, correcting hallucinations.
• Focus on making knowledge executable for both people and machines.

Key takeaway: Writers must stop competing with AI and instead collaborate with it: surfing the wave by focusing on pedagogy, empathy, and context, while letting AI handle boilerplate and first drafts.