# The Chief AI Officer: Scientist, Architect, Coach

Rania Khalaf, WSO2 | AI Engineer | 22:22

Source: https://www.youtube.com/watch?v=9cJrbj23fOA
Channel: AI Engineer (https://www.youtube.com/@aiDotEngineer). Summarised by AIE Talks.
Page: https://aietalks.com/talks/the-chief-ai-officer-scientist-architect-coach
Published: 2026-09-30
Tags: enterprise, product-strategy, team-adoption

## TL;DR
- The Chief AI Officer role changes with the company, its AI maturity, and the person's own background.
- Khalaf divides the role into Scientist, Architect, and Coach, with each area set at a different level depending on the situation.
- She measures AI fluency, adoption, product readiness, market signals, and AI revenue instead of counting tokens.

## Summary
Rania Khalaf describes the Chief AI Officer as a fluid role that can mean very different things across companies. She proposes three areas of focus: Scientist, Architect, and Coach. Each is a slider. A scientist experiments and builds, an architect shapes business and product strategy, and a coach helps employees, customers, and communities adopt AI. Khalaf compares her work at IBM Research, an agricultural biotech company, and WSO2 to show how the balance changes. At WSO2, she spends about 20% of her time as a scientist, 60% as an architect, and 20% as a coach. She rejects token counts as a measure of AI progress. Instead, she tracks workforce fluency, adoption depth, agent-consumable products, pricing that can handle agent usage, earned thought leadership, community activity, and AI revenue. She also argues that the role needs strong CEO support and may belong to a CPO or head of HR in some companies.

## Key ideas
### The role changes with the company and the person
[01:09](https://www.youtube.com/watch?v=9cJrbj23fOA&t=69s)
Khalaf says the Chief AI Officer role has different meanings depending on the type of company and its level of AI maturity. A third factor is the person's own background and skills. Some companies create the role for a particular person, while others start with a specific business need and then hire someone to address it. Because the role is difficult to hire for, companies often stay open-minded about what the person will do. Khalaf has also seen the job split into two roles when one person cannot cover the full range.

### Three areas keep an overloaded role manageable
[02:57](https://www.youtube.com/watch?v=9cJrbj23fOA&t=177s)
Khalaf divides the job into Scientist, Architect, and Coach. The Scientist explores, experiments, and builds. The Architect also builds, but takes responsibility for strategy, product direction, new revenue, and the company's wider use of AI. The Coach educates employees, customers, and communities. In a board-facing role, she also reports how the company's AI work helps the business and its customers. Her coaching includes trusted-advisor conversations with customers, where she starts with their problems and may recommend another technology if WSO2 is not the right fit.

### Each area is a slider rather than a fixed job description
[05:38](https://www.youtube.com/watch?v=9cJrbj23fOA&t=338s)
The balance between the three areas depends on the company and the person's skills. A Scientist may mainly explore new technology, while the other end of that slider involves inventing and creating. An Architect might manage a budget, prioritize proposed projects, and report on them without deep AI knowledge. At the other end, the Architect shapes company strategy, productivity, products, and go-to-market work. Coaching also changes with maturity. Less mature companies need more internal education, while mature companies may spend more time working with customers and the community.

### Khalaf's career shows why the sliders move
[08:32](https://www.youtube.com/watch?v=9cJrbj23fOA&t=512s)
After 20 years in IBM Research, Khalaf joined an agricultural biotech company working with CRISPR on corn, soy, and wheat. The company initially wanted help with AI and data, but her mandate expanded to include engineering, data infrastructure, and hiring a CISO. The work involved AWS, Databricks, data cleanup, and gene-discovery algorithms, alongside basic operations. At WSO2, she estimates her balance at 20% Scientist, 60% Architect, and 20% Coach. She spends more time shaping the company's strategy around the agentic enterprise and coaching outward because WSO2 has a highly technical workforce.

### Simple computer vision can be better than machine learning
[11:12](https://www.youtube.com/watch?v=9cJrbj23fOA&t=672s)
In the biotech company, scientists wanted to measure the size of a corn embryo because it indicated how likely the next gene-editing step was to succeed. Human measurement was too expensive, so the team considered AI. Khalaf found that the problem did not require machine learning. Basic blob detection was enough, and the simple algorithm worked well. She uses this example to show that the right AI solution may be a known technique applied to a clear operational problem rather than a new model.

### AI progress should be measured through capability and adoption
[12:52](https://www.youtube.com/watch?v=9cJrbj23fOA&t=772s)
Khalaf refuses to measure tokens because the number can be easily manipulated. She looks at AI fluency across the workforce, including whether teams can build and use AI without relying on a small central group. She tracks adoption depth, from simple tasks such as email cleaning or code completion to changes in whole workflows and the use of agentic employees. She also tracks tool availability and GEO visibility, meaning whether language models can find and understand the company. The measures should change as the company and its AI work mature.

### Products need to be usable by agents and priced for agent usage
[15:02](https://www.youtube.com/watch?v=9cJrbj23fOA&t=902s)
Khalaf says WSO2 wanted its products to be consumable by agents and language models. The company made documentation language-model consumable, added support for agents, language models, and tools, and is working so every product can be used by agents. That includes MCP servers, skills, and CLIs. She also discusses agent-proof pricing. If agents drive much higher consumption, pricing based on seats may stop working, so WSO2 uses consumption-based pricing. She wants the company to use its own products and listen to feedback from that use.

### The role needs CEO backing and may belong elsewhere
[20:01](https://www.youtube.com/watch?v=9cJrbj23fOA&t=1201s)
Khalaf says there is no single structure for the role. A Chief AI Officer needs a strong relationship with the CEO because the work crosses many parts of the company. In an AI-forward software company, the Chief Product Officer and Chief AI Officer may be the same person. In companies that do not sell software, she has seen the head of HR take responsibility for AI because agents are treated as part of the workforce. Her final advice is to shape the role around what a person does well, what they enjoy, what the company needs, and what they can be paid for.

## Notable quotes
- "The tricky thing about this role is like it's so great. Okay, I have an AI officer, but there's AI in everything, right?" (02:08)
- "I use the word scientist, architect, and coach." (02:57)
- "I was like, well, yeah, sure, we can use AI, but you just need like blob detection. You don't need any machine learning." (12:05)
- "You really get what you measure. You start measuring something, everyone's going to optimize for that." (17:34)
- "There's no silver bullet. It depends on the company." (19:59)

## Tools & references mentioned
- IBM Research
- WSO2
- Claude
- Gemini
- AWS
- Databricks
- CRISPR
- MCP servers
- agent identity
- AI gateway
- agent builder
- agent manager

## Who should watch
- You are defining a Chief AI Officer role and need a practical way to decide how much of the job should involve research, strategy, or education.
- Your company is adopting AI across products and teams, and you need measures that go beyond token volume or isolated experiments.
- You are deciding whether AI ownership belongs with a dedicated executive, the CPO, the CTO, or another business leader.

## Related talks

- [Rise of the AI Architect](https://aietalks.com/talks/rise-of-the-ai-architect) (Clay Bavor, Sierra & Alessio Fanelli, Latent Space, 18:55)
- [Agentic Enterprise: What Your CEO Must Know About AI](https://aietalks.com/talks/agentic-enterprise-what-your-ceo-must-know-about-ai) (Hubert Misztela, Novartis, 28:04)
- [Leadership in AI Assisted Engineering](https://aietalks.com/talks/leadership-in-ai-assisted-engineering) (Justin Reock, DX, 18:11)
- [Opening Keynotes - AIE Paris 2025 (Day 1)](https://aietalks.com/talks/opening-keynotes-aie-paris-2025-day-1) (swyx, Latent Space and AI Engineer & Lélio Renard Lavaud, Mistral AI, 1:25:26)
- [Structuring a Modern AI Team](https://aietalks.com/talks/structuring-a-modern-ai-team) (Denys Linkov, Wisedocs, 17:40)