In the third installment of the six-part series: “Building the Next-Gen CoE for the Age of AI Agents,” Jon Tudor, Director of Business Architecture at Dataiku, and Burdell Schwartz, Regional Vice President of Dataiku, build a compelling roadmap for organizations seeking to mature their AI and analytics capabilities. At the heart of this discussion is a deeply practical question: How can organizations operationalize CoEs in a way that scales self-service, empowers people, delivers measurable value, and embeds AI into the fabric of business operations?
Jon and Burdell walk through a comprehensive framework that touches every layer of this challenge — from team design and innovation methodology, to support models, funding strategies, and data product governance. Each section of this webinar recap logically feeds into the next, painting a clear picture of how organizations can go from concept to scaled capability, all while leveraging the power of The Universal AI Platform™.
Building a Self-Service Team: Foundations for Scale
Jon begins by laying out the cornerstone of any successful CoE: an intentionally structured team. The goal? To move from centralized control to a model where enablement is embedded in the business, making it easier for teams across functions to adopt AI independently and at scale. To achieve this, Jon explains, organizations need to think in three key team vectors:
1. Core CoE Team
This is the operational hub of your CoE. These individuals own the day-to-day and strategic direction:
- Product Owner: The architect of vision and backlog. They engage across the business to capture needs and translate them into prioritized work.
- Platform Architect: Oversees governance, scalability, and platform design to ensure Dataiku runs efficiently and securely.
- Applied SME: Domain experts (e.g., data scientists, engineers) who coach users and support AI and AI agent adoption through subject matter knowledge.
2. Extended Team
This group scales impact horizontally across the organization:
- Trainers and Outreach Leads: Upskill teams and promote a culture of AI.
- Operational Support: Provide hands-on assistance and help users navigate both the platform and use cases.
3. Business Team
The embedded champions who bridge grassroots needs with enterprise strategy:
- Ambassadors: Power users who promote adoption and give feedback.
- Champions: Senior executives who advocate for AI and AI agents within their business domains, shaping priorities and allocating resources.
Together, these team vectors create a flexible structure that promotes ownership without losing control — a core requirement for scale.
Driving Internal Innovation Through Design Thinking
With the team in place, the next question becomes: How do we move from enablement to innovation? Jon points out that true transformation comes not from tools, but from helping the business solve real problems, faster.
He introduces design thinking as the preferred approach. It enables organizations to align with user needs, create better data products, and surface innovations from within the business.
The Five Stages of Design Thinking
- Empathize: Observe users in context; understand their goals and blockers.
- Define: Clearly articulate their core challenges.
- Ideate: Brainstorm creative ways to reduce friction.
- Prototype: Rapidly test concepts.
- Test: Validate at scale and iterate.
By applying Agile principles throughout the design thinking cycle, teams can introduce structure and repeatability into the innovation process — making it easier to move from ideas to execution.
Equally important is the need to observe how users actually work, rather than relying solely on what they say. Jon highlights the role of ambassadors in this context: they’re embedded in the business, close to real-world use cases, and ideally positioned to test prototypes early and provide fast, meaningful feedback.
From Break-Fix to Consultative Support
Innovation and enablement must be sustained. And that requires a shift in support philosophy. Traditional IT models focus on uptime and issue resolution. But in a CoE-led, self-service environment, support becomes something more: a coaching function.
Jon outlines several pillars to evolve support:
- Office Hours: Open, unstructured sessions once or twice a week encourage cross-team dialogue, reduce friction, and foster community.
- Responsive Help: Quick responses via Slack or chat tools reduce user frustration. Tickets can be logged post-resolution.
- Tiered Strategy: Match support to complexity — use documentation and chatbots for basic issues, reserve SMEs for deeper challenges.
- Data-Driven Prioritization: Use support and usage data to uncover the "20% of problems causing 80% of pain."
- Unified Entry Point: As the program matures, combine support for AI, data infra, and visualization into a single gateway.
This consultative model not only improves user satisfaction — it enables the CoE to scale without burning out.
If the organization gets requests for a hundred data products to be executed, but only 10 can be, the other 90 need a place to go. That’s where self-service comes into the frame.
- Jon Tudor, Director of Business Architecture, Dataiku
Despite all the right structures, centralized teams will always hit capacity. As Jon explains, self-service isn’t just a convenience — it’s a strategic necessity. It gives teams across the business a way to prototype, test, and deliver value even when their projects aren’t prioritized centrally. And to manage that growing body of decentralized work? That’s where program management metrics come in.
Using Program Management Metrics to Drive Strategy
As CoEs evolve, measurement becomes mission-critical. Jon emphasizes that metrics aren’t just for tracking success, they guide strategic direction. He encourages starting small:
- Choose two or three key metrics aligned with your core goals.
- Examples include ROI of AI initiatives, Net Promoter Score, feature adoption, and asset reuse.
Jon recommends ensuring your metrics are:
- SMART: Specific, measurable, achievable, relevant, and time-bound — to ensure they are actionable and meaningful.
- Automated: Data collection should be seamless and built into workflows to support sustainability.
- Centralized: Metrics should be consolidated and observable across teams for consistent visibility.
- Actionable: The metrics must inform decisions, shape backlog priorities, and support strategic conversations at the executive level.
He also encourages regular evaluation: What worked 12 months ago may not reflect current strategy or platform maturity.
Dataiku Global Services: Accelerating Operationalization at Scale
To close the session, Burdell Schwartz walks through how Dataiku’s Global Services can help organizations bring these ideas to life. Operationalizing a CoE doesn’t just require vision, it also requires dedicated roles to support and guide execution.
Burdell introduces two key roles:
1. Technical Account Manager (TAM)
- Experts in administration, architecture, and performance optimization
- Advisors on platform tuning, configuration, and governance
- Mentors for in-house admin teams to build long-term self-sufficiency
2. Data Scientist in Residence (DSIR)
- Embedded experts in data science and analytics
- Partners in ideation, prototyping, and operationalization
[Technical Account Managers and Data Scientists in Residence] are not just solving problems — they’re also building capabilities within your organization and fostering a culture of self-sufficiency.
- Burdell Schwartz, Regional Vice President of Services, Dataiku