Agent demos are everywhere. Production ROI is not. The gap isn’t the model, the framework, or the prompt, it’s everything below the waterline: developer experience, agent identity, and inference economics.

In this on-demand session, originally hosted with ODSC, DataRobot’s Nikki Rouda breaks down the three engineering foundations that decide whether agents deliver real ROI in production — and how to make enterprise readiness a platform capability instead of a project.

What you’ll learn

• Why 80% of engineering effort goes to the 30% nobody puts in the demo, and how to close the gap without rewriting your agent code
• How to give agents first-class identities with scoped, auditable authority, instead of shipping on shared API keys
• How to keep inference costs predictable at scale with token pools and premium/spot tiers
• Key findings from new DataRobot research with 413 AI leaders on what blocks agentic AI ROI

Who should watch

• CIO, CTO, Head of AI
• Chief Data Officer
• VPs of AI/ML, Data Science, Platform Engineering
• ML Engineers, AI Architects