DataRobot

Building an AI agent demo is easy. Getting it into production is where things get hard. In this talk, Amber Bennoui, Developer Relations lead at DataRobot, breaks down the gap between “I built an agent” and “my company is actually using it.” The real blockers are not the model, the framework, or the prompt. They are the production layers most teams ignore: developer experience, observability, agent identity, authorization, governance, cost control, and scalable infrastructure. You’ll learn why agents need to be treated like first-class identities, how to avoid API key sprawl, why evals and regression testing matter, and how token economics can make or break your production strategy. If you are building agents for real enterprise use cases, this is the checklist you need before you ship. Topics covered: AI agents in production Agent identity and authorization DevX for agent builders Observability and evals Token economics and infrastructure Enterprise AI governance Moving from demo to production Stay connected with DataRobot! 👉 Website: https://www.datarobot.com/ 👉 LinkedIn: https://www.linkedin.com/company/datarobot/ 👉 Blog: https://blog.datarobot.com/

Your agent framework isn’t the bottleneck. The bottleneck is every new tool, wrapper, and MCP server your agent needs before it can do useful work. In this Build Club session, Luke Shulman, Director of AI Innovation at DataRobot shows how to build an agent that writes its own tools from an OpenAPI spec, runs them in a secure sandbox, and turns the working tools into a deployable agent artifact. The live example tackles CODEOWNERS cleanup in a real monorepo: finding stale ownership references, proposing fixes, and keeping a human in the loop before anything ships. What you’ll learn: - Why curated tool registries don’t scale - How agents can use OpenAPI specs as context - Why the sandbox matters more than the agent loop - How generated tools become part of a production-ready artifact 🔗 Read the full blog: https://www.datarobot.com/blog/agent-that-writes-its-own-tools/ 🔗 NL agent runtime: https://github.com/kindofluke/context-agent 🔗 DataRobot Agent Skills: https://github.com/datarobot-oss/datarobot-agent-skills Stay connected with DataRobot! 👉 Website: https://www.datarobot.com/ 👉 LinkedIn: https://www.linkedin.com/company/datarobot/ 👉 Blog: https://blog.datarobot.com/ #AIagents #AgenticAI #DataRobot #OpenAPI #MLOps #DeveloperTools