Production AI Agents & AI Workflows with Hopsworks: Build & Deploy, Live Demo

Every AI agent looks good in a demo. Not every AI agent survives production. 

The gap between a working prototype and a reliable production agentic system is rarely an agent problem. It is an infrastructure problem. Agents need to be connected to one or more LLMs, data sources for context, a compute environment for running the agent, and infrastructure for tracing, monitoring, and development. This is even more of a challenge in sovereign, air-gapped and on-premises scenarios.

In this 30-minute live session, we'll close that gap. We'll build and deploy a real AI agent using Hopsworks 5.0, so you can see exactly what production-grade agentic systems are actually built on. Even in air-gapped environments.

What we'll cover:

• Batch agent harnesses for AI workflows: when to use a mature agent harness like Claude Code or Codex versus building your own.
• Real-time agent harnesses for agent deployments: how to build on open-source frameworks like LlamaIndex, use OTEL for tracing and evals, and how to integrate real-time context.
• LLM serving in air-gapped environments with Hopsworks and connection with coding agents in Hopsworks (Claude Code or OpenCode)
• What's new in Hopsworks 5.0: ai agents as deployments with LlamaIndex, OTEL traces, and AI workflows with claude code, codex, and more

Who this is for: ML engineers, data engineers, and platform teams building agentic systems.

📅 June 25, 2026 · 9 AM EDT | 3 PM CEST · 30 minutes · Free