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.
Summary
How can retrieval-augmented generation systems be designed to correctly distinguish between personal data queries, product information requests, and general conversation, ensuring accurate answers and protecting user trust?
In his presentation, Selvaraaju Murugesan (Head of Data Science at Kovai.co) explores how technical writers can adapt content for AI consumption, as documentation audiences shift from “humans only” to “humans plus bots.” With LLMs like ChatGPT and Claude increasingly ingesting and reasoning over docs, priorities move from SEO and keywords toward semantic clarity and structured content.
Writers are positioned as “truthkeepers”: since AI agents can act on documentation (e.g., executing procedures), content must be precise, consistent, and free of ambiguity. Best practices include comprehensive detail, well-structured FAQs, consistent terminology, explicit headings, and bias-free language. Common pitfalls—like generic or duplicate titles, poor glossary structure, or overly long procedures that exceed token limits—can break chatbot performance.
To ensure reliability, teams should evaluate chatbots before launch: define test cases, establish ground-truth answers, measure accuracy with open-source frameworks, and fine-tune content and prompts.
Takeaway: AI-ready documentation requires rethinking content strategy, writers now author for both people and machines, safeguarding trust at scale.
Sign up to our Developer Portal Newsletter so that you never miss out on the latest API The Docs recaps and our devportal, API documentation and Developer Experience research publications.
