This blog recap summarizes the key takeaways from the first webinar in Jon Tudor’s six-part series: “Building the Next-Gen CoE for the Age of AI Agents.” As Director of Business Architecture at Dataiku, Jon explores how centers of excellence (CoEs) can enable scalable, enterprise-wide self-service in the age of AI agents — while staying tightly aligned to business outcomes and operational realities.
Jon draws from his experience leading a large-scale self-service and data governance program at GE Aerospace, and from working directly with 80+ Dataiku customers. The result: a clear-eyed, user-first approach to structuring programs that work.
This session is about building systems that reflect how people work. And how CoEs can lead that change.
The Landscape Has Shifted — and CoEs Must Shift With It
AI agents are being adopted faster than expected. By 2025, 25% of enterprises are expected to deploy them, growing to 50% by 2027. But as Jon points out, the most important shift isn’t technical — it’s operational and cultural. Most organizations don’t suffer from a lack of AI capability. They struggle because access to data is still uneven, workflows are fragmented, and teams lack clarity on how to engage with AI.
Meanwhile, the nature of work is changing. In 1960, decision-making jobs accounted for just six percent of the workforce. Today, that number is over 34% — and growing. According to the International Monetary Fund, up to 60% of jobs in advanced economies are expected to be reshaped or replaced by automation.
That’s why CoEs can’t just maintain systems. They need to lead the organization into a new way of working — with AI as a partner, not an afterthought.
Strategy Begins With Listening
Too often, self-service efforts begin with tooling. Jon argues that they should begin with listening.
The foundation of any effective CoE program is a deep understanding of the people you’re enabling. Through structured interviews, Jon helps organizations learn how users currently engage with data, what gets in their way, and where opportunities exist for AI to help them work smarter.
This discovery process uncovers:
- Skill levels and comfort with data tools (from spreadsheets to prompt engineering)
- Current pain points (access, complexity, duplication of work)
- Business outcomes they’re aiming for — and what’s blocking them
- Frustrations and wish lists
- Whether they might serve as future champions or power users
As someone who spent over a decade at GE Aerospace leading large-scale data programs, I’ve seen firsthand how the right structure — not just tools — makes or breaks a self-service initiative.
- Jon Tudor, Director of Business Architecture at Dataiku
These interviews also help define your total addressable market for self-service: who should be enabled, who shouldn’t, and which AI use cases are worth pursuing.
From Conversations to Personas: Structuring Around Real Users
Once the interviews surface patterns, Jon recommends creating personas — realistic representations of key user types, based on skills, behaviors, and team needs.
At GE Aerospace, his team brought these personas to life using Star Wars archetypes to make them memorable across the business:
- Han Solo: Business veterans with deep domain expertise but limited exposure to data tools. Often stuck in spreadsheets. They need help moving toward structured, reusable pipelines.
- Rey: Technically curious, often early in their careers, and already experimenting with low-code or scripting. They’re well-positioned to bridge business and technical teams.
- Yoda: Experienced data scientists or engineers. They don’t need training — they need systems that support fast iteration and experimentation.
These personas shaped every decision — who gets access to what, how training is delivered, and what kinds of AI agent use cases make sense. In many organizations, this approach has scaled to 30+ personas that underpin entire data literacy programs.
Personas allow you to meet people where they are — and build systems around how they actually work.
The 6S Framework: A Model to Guide the Work
Personas help define who you’re enabling. But you also need a way to define how self-service work gets done. That’s where Jon introduces the 6S Framework — a practical, repeatable model used by many Dataiku customers (and previously at GE Aerospace) to support users at every stage.
Each “S” represents a step in the data-to-decision journey:
- Search: Can users find what already exists — datasets, models, prior projects? If not, they’ll rebuild.
- Stitch: Can they prepare and join data from different sources, without needing custom SQL or engineering support?
- Science: Can they build models or run predictions — regardless of their coding background?
- Synthesize: Can they apply LLMs or AI agents to automate outputs or actions?
- Show: Can they visualize the outcomes clearly — whether in dashboards or other formats — and drive decisions?
- Share: Can they document, reuse, and scale their work across the organization>
This framework simplifies complex work. It creates a shared language between CoEs and the teams they serve. And it makes your program repeatable — across use cases, departments, and user types.
Jon also highlights how organizations can map each “S” to their tech stack. For example, within Dataiku, teams can complete all six steps using visual recipes or code. But the model is tool-agnostic: Every company should tailor it to their architecture and make it easy for users to follow.
Designing for the Future of Work
Jon closes with a simple truth: Scaling self-service isn’t about handing out tools and hoping for the best. It’s about designing systems — grounded in structure and clarity — that reflect how people work today, and how they’ll work tomorrow.
The CoE’s role in this environment is to enable, not gatekeep. To guide, not dictate. This first session lays the foundation for what’s ahead. In the next five webinars, Jon will dive deeper into governance, organizational structure, architecture, and change management — always rooted in real-world examples from leading Dataiku customers.
If you're shaping how your company works with data and AI, this series is your blueprint. Each session builds on this foundation — giving CoEs the frameworks, language, and playbooks they need to lead in the AI-powered enterprise.