Nick Gomez (Inkeep): Using AI to Drive Content Roadmap and Documentation Style

Nick's presentation

In his presentation, Nick Gomez discusses how integrating AI tools like chatbots and search co-pilots into documentation can enhance user experience and support. He highlights key learnings from their extensive experience handling high volumes of queries and offers insights into best practices for content creation and optimization for AI interactions. 

Key Takeaways

• Dual Audience Consideration 
  • Users and AI Interaction: Documentation should cater not only to human users but also to AI models. AI uses the same content to interact with services and answer user queries. Therefore, ensuring clarity in documentation benefits both human and AI audiences. 
• Importance of Open API Specifications 
  • Detailed API Documentation: High-quality Open API specifications are crucial as AI models use these to interact with services. Clear naming, descriptions, and examples help AI models understand and utilize APIs effectively. 
• Effective Content Design 
  • Hierarchical Structure: Organize content hierarchically to provide context for AI models. Structured documentation with clear headings and navigation helps AI systems understand and retrieve information accurately. 
  • Avoid Monolithic Pages and Client-Side Loaded Content: Break down lengthy pages and avoid content that loads dynamically via client-side scripting to improve indexing and retrieval by AI systems. 
• Addressing AI Retrieval Challenges 
  • Hallucinations vs. Conflations: Distinguish between hallucinations (AI-generated incorrect information) and conflations (misinterpretations due to lack of context). Use retrieval-augmented generation (RAG) to ground AI responses in factual content and provide accurate answers. 
  • Testing and Iteration: Regularly test AI systems with known fail cases and iterate based on feedback. Create a "golden set" of questions to evaluate and refine AI performance. 
• Optimizing Content for AI 
  • Scenario-Based Content: Include scenario-focused examples in documentation to match user queries with relevant solutions. This helps AI systems better understand and respond to practical use cases. 
  • Disambiguation and Examples: Clearly define similar concepts and provide end-to-end examples to facilitate AI comprehension and accurate responses. 
  • Feedback Loop: Use insights from user interactions with chatbots to drive content improvements and identify gaps in documentation. 
• Using AI for Content Improvement 
  • Analyzing User Queries: Leverage tools like Inep to analyze chat interactions and generate reports on documentation gaps and feature requests. This helps in continuously refining content based on real user needs. 

Nick concludes by demonstrating Inep’s capabilities in monitoring and reporting on chatbot interactions, which aid in identifying documentation gaps and improving overall content quality. The emphasis is on iterative development and leveraging AI to enhance both user and AI experiences with documentation.