> Markdown version of [/videos/100284-your-docs-are-now-ai-infrastructure-treat-them-like-it?t=3](https://www.wearedevelopers.com/videos/100284-your-docs-are-now-ai-infrastructure-treat-them-like-it?t=3). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Your Docs Are Now AI Infrastructure (Treat Them Like It)

Are outdated documents sabotaging your AI tools? Autonomous agents are now the primary readers of your technical content. Treat docs as vital infrastructure to unlock flawless user experiences.

- **Speakers:** [Emil Sorensen](https://www.wearedevelopers.com/@emil-sorensen)
- **Event:** World Congress 2026 Europe
- **Published:** July 10, 2026
- **Duration:** 9:29
- **URL:** https://www.wearedevelopers.com/videos/100284-your-docs-are-now-ai-infrastructure-treat-them-like-it

## Summary

Software documentation has quietly evolved from a static reference for developers and support teams into essential grounding infrastructure for artificial intelligence. With the rise of tools like Claude Code, Codex, Cursor, and custom in-product sidekicks, autonomous agents are increasingly the primary consumers of technical content. If documentation is messy, outdated, or incomplete, the AI directly inherits those blind spots, significantly degrading the user experience. By analyzing traces from real-world agent deployments involving complex workflow tasks, kapa.ai discovered that knowledge-base search is the single most dominant tool called by agents, utilized in over a third of all interactions. Users naturally treat embedded agents as universal support bots, forcing the AI to rely heavily on structured documentation to bridge the gap between user intent and platform capabilities. Treating docs as code and AI infrastructure creates powerful interaction effects. When users issue poorly phrased queries or ask for native actions—such as explaining model context protocol integrations or analyzing negative sentiment—agents frequently use documentation search as a fallback reasoning loop to plan and properly execute their tasks. Leveraging high-quality retrieval-augmented generation systems allows engineering and product teams to uncover the dark matter of documentation, revealing exactly which topics generate uncertainty. Instead of guessing what to write, teams can utilize aggregated agent queries to prioritize updates, ultimately achieving a 20 to 40 percent deflection in support tickets and saving hundreds of developer days.

**Keywords:** ai agent infrastructure, support ticket deflection, knowledge base optimization, retrieval-augmented generation, workflow agents, in-product ai sidekicks, developer documentation strategy, kapa.ai, claude code, model context protocol integrations, ai tool call analytics, documentation knowledge engine, customer-facing ai chat, copilot deployments

## Chapters

1. **The emergent readers of software documentation** (00:03) — The rise of AI agents like Claude Code shifts documentation audiences from humans to automated systems.
1. **Classifying agents that consume documentation** (01:12) — How in-product copilots and workflow agents interoperate with existing documentation alongside human readers.
1. **Tool call dominance in production AI agents** (03:29) — Analyzing internal agent data reveals knowledge base search as the most utilized tool over native product actions.
1. **The interplay of native actions and knowledge search** (05:59) — Agents frequently query documentation to interpret the results of their native tool execution properly.
1. **Utilizing documentation to plan ambiguous user queries** (07:12) — Agents lean on knowledge base search to contextualize poorly formulated requests and find proxy solutions.
1. **Ensuring retrieval quality for knowledge base search** (08:31) — Why treating docs as AI infrastructure requires robust retrieval systems over basic web search APIs.

## Related Moments

- [Treating developer documentation as a primary product](https://www.wearedevelopers.com/videos/1801-api-first-how-twilio-designs-for-developers-justin-kitagawa-twilio) (from "API‑First: How Twilio Designs for Developers - Justin Kitagawa (Twilio)")
- [Transitioning documentation from human readers to AI agents](https://www.wearedevelopers.com/videos/100199-making-documentation-ai-ready-preparing-your-docs-for-the-llm-era) (from "Making Documentation AI-Ready: Preparing Your Docs for the LLM Era")
- [Adapting developer relations strategies for agent engine optimization](https://www.wearedevelopers.com/videos/1898-developer-relations-in-the-world-of-agents-michelle-duke) (from "Developer Relations in the World of Agents - Michelle Duke")
- [Deploying contextual documentation bots to accelerate developer learning](https://www.wearedevelopers.com/videos/1267-exploring-ai-opportunities-and-risks-in-development) (from "Exploring AI: Opportunities and Risks in Development")
- [Prioritizing human readable documentation despite AI context indexing](https://www.wearedevelopers.com/videos/1346-wearedevelopers-live-blockchain-after-the-hype-vibing-all-the-things-big-tech-and-work-best-practices-more) (from "WeAreDevelopers LIVE - Blockchain after the hype, Vibing all the Things, Big Tech and Work Best Practices & more")
- [Addressing AI assistance and documentation tooling](https://www.wearedevelopers.com/videos/681-technical-documentation-how-can-i-write-them-better-and-why-should-i-care) (from "Technical Documentation - How Can I Write Them Better and Why Should I Care?")

## Related Articles

- [Humanizing Your Documentation](https://www.wearedevelopers.com/magazine/133-humanizing-your-documentation)
- [Never delegate the understanding](https://www.wearedevelopers.com/magazine/749-never-delegate-the-understanding)
- [The real reason we document our code](https://www.wearedevelopers.com/magazine/518-the-real-reason-we-document-our-code)
- [Graph and AI Trends 2026: Why Is AI Running but Not Yet Delivering?](https://www.wearedevelopers.com/magazine/680-graph-and-ai-trends-2026-why-is-ai-running-but-not-yet-delivering)

## Related Jobs

- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [Principal Product Manager, Agent Platform](https://www.wearedevelopers.com/jobs/ext/277541-principal-product-manager-agent-platform) at **GitHub**
- [Manager, Technical Support](https://www.wearedevelopers.com/jobs/ext/737543-manager-technical-support) at **Twilio**
- [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace**
- [Principal Field Architect - AI Agents](https://www.wearedevelopers.com/jobs/ext/1442858-principal-field-architect-ai-agents) at **Twilio**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**