{ "video": { "id": "9cJrbj23fOA", "title": "The Chief AI Officer: Scientist, Architect, Coach — Rania Khalaf, WSO2", "duration": 1342, "upload_date": null, "channel": "AI Engineer", "source": "AI Engineer", "event_format": "" }, "analysis": { "video_id": "9cJrbj23fOA", "title": "The Chief AI Officer: Scientist, Architect, Coach — Rania Khalaf, WSO2", "one_liner": "Rania Khalaf, Chief AI Officer at WSO2, describes the role as three focus areas (scientist, architect and coach) whose balance depends on whether the company sells software, how technical its workforce is and how mature its AI is. She refuses to measure AI by tokens and tracks fluency, adoption depth, agent-readiness and earned signals instead.", "summary": "Khalaf argues that the Chief AI Officer role means very different things depending on the company type, its AI maturity and the background of the person in the role. To stay focused, she splits the job into scientist (exploring and experimenting), architect (strategy, product and the board-facing parts of the job) and coach (evangelizing, educating and acting as a trusted advisor, inside or outside the company), each running along a range. She compares her time at a CRISPR agriculture biotech, where the mandate grew to cover IT, security and data, with WSO2, where she is about 20% scientist, 60% architect and 20% coach. She also explains what she measures and why tokens are a bad metric.", "key_points": [ "She cites IBM studies showing companies with a Chief AI Officer rising from 11% in 2024 to 26% in 2025 and now 76%.", "The role varies by company type and AI maturity, and it is often shaped around whoever is hired. Some CAIOs are business-savvy stewards who prioritize projects, allocate budget and report out, with little deep AI knowledge. Others, like her, reshape product strategy and go-to-market.", "Coaching was the part that surprised her after 20 years at IBM Research. It takes a lot of evangelizing and overcommunicating. Less technical companies need inward coaching, where people have to experience hallucination for themselves before they believe it. More mature companies need outward coaching of customers and the community. At WSO2, where about 75% of staff are technical, she coaches mostly outward.", "At the agriculture biotech, which used CRISPR on corn, soy and wheat, her remit grew to cover IT and hiring a CISO. She moved the company to AWS, bought Databricks and migrated it from Box to Microsoft. When scientists asked for AI to measure the size of corn embryos, she used simple blob detection rather than machine learning, and it worked well.", "She refused a request to measure token usage because tokens are 'so easily hackable'. Instead she tracks AI fluency across the workforce, builders across functions, level of adoption (from email cleanup and code completion up to 'agentic employees'), tool availability through team or enterprise accounts, and GEO (generative engine optimization) visibility.", "For software products she tracks: docs that LLMs can consume; first-class support for agents, LLMs and tools (added last year); every product usable by agents through an MCP server, skills and a CLI (this year); 'agent-proof' consumption-based pricing; and dogfooding with a closed feedback loop.", "For market and ecosystem she tracks earned media (not paid articles), analyst recognition, customer adoption, open-source community signals (WSO2 is 100% open source, not open core, so stars, forks and adoption count) and ARR from AI products. She says the list should change as these areas mature.", "WSO2 is about 20 years old with around $150M ARR. It extended its API, integration, identity and internal-developer platforms with an AI gateway, agent identity, an agent builder, an agent platform and agent manager. Its small research team published three papers this year, two with undergraduates in Sri Lanka, one of which won a best paper award." ], "takeaways": [ "Split the CAIO job explicitly into scientist, architect and coach, and decide where you sit on each range for your company type, workforce and AI maturity.", "Don't use token usage as your AI metric. Measure fluency, builders across functions, depth of adoption, tool access and external signals, and change the metrics as each one matures, because 'you get what you measure'.", "Make products agent-consumable: docs LLMs can read, an MCP server, skills and a CLI for every product, and consumption-based pricing that survives agent-driven usage.", "Keep part of the team always experimenting with new AI tech, and reach for the simplest technique that works (for example blob detection instead of ML).", "Get strong CEO backing before taking the role, and shape it around what you're good at, what you love, what the world needs and what you can be paid for." ], "topics": [ "ai leadership", "chief ai officer", "ai strategy", "metrics", "agents", "ai adoption", "generative engine optimization", "open source" ], "tools": [ "WSO2", "IBM", "Claude", "Gemini", "AWS", "Databricks", "Box", "Microsoft", "CRISPR", "MCP" ], "quotes": [ { "text": "You want them to discover for themselves things like hallucination because you can explain until you're blue in the face, but people don't believe it until they experience it.", "at": "07:30", "url": "https://www.youtube.com/watch?v=9cJrbj23fOA&t=450s" }, { "text": "well yeah sure we can use AI but you just need like blob detection you don't need any machine learning", "at": "12:05", "url": "https://www.youtube.com/watch?v=9cJrbj23fOA&t=725s" }, { "text": "you really get what you measure. You start measuring something, everyone's going to optimize for that.", "at": "17:34", "url": "https://www.youtube.com/watch?v=9cJrbj23fOA&t=1054s" }, { "text": "So everything needs to have an MCP server and skills and CLI etc.", "at": "15:35", "url": "https://www.youtube.com/watch?v=9cJrbj23fOA&t=935s" } ], "words": 4519 }, "summary_url": "/#9cJrbj23fOA", "transcript": { "html": "/transcripts/9cJrbj23fOA.html", "txt": "/transcripts/9cJrbj23fOA.txt", "vtt": "/transcripts/9cJrbj23fOA.vtt" } }