The New Wave of Agent Architectures: Claude Agent SDK & What’s Actually Working in Production

​Everyone is building agents. Few are running them in production.

​Claude’s Agent SDK and evolving OpenAI tooling have made it easier than ever to prototype. But once you move from demo → workflow, things get complicated fast.

​What holds up?
What breaks?
What scales?

​AI adoption is moving fastest inside AI-native, high-growth companies and engineers are driving that shift.

​This roundtable is for builders who care about the difference.

​We’re excited to dig into:

• ​How teams are architecting agents with Claude’s Agent SDK

• ​What fails when you move from playground to production

• ​Which engineering challenges in agent systems are sticky

​We'll frame the discussion through the lens of building legal-grade AI systems, where reliability, evaluation, orchestration, and workflow integration create real constraints.

​Who It’s For
Engineers & PMs who are actively building AI systems in startup or high-growth environments.

​We’re keeping the room intentionally small to maintain depth and conversation.

​Expect good food, drinks and a lively discussion with a room full of builders.