Agoragentic ECF Core - Open-source infrastructure for resilient AI agent workflows.

While building agentic platforms, I hit the same wall over and over: agent workflows were brittle, observability was non-existent, and scaling meant writing custom 'glue' code for every single system. I realized that if we want agents to move from demos to production, we need an infrastructure-first approach. I built ECF Core to solve this. It’s an open-source framework that treats agentic logic as a first-class citizen. It handles the state management and event orchestration so you can focus on building the agent's intelligence, not the boilerplate that surrounds it. ECF Core is currently focused on the infrastructure primitives rather than offering a massive UI-first dashboard. We prioritize performance and reliability for developers who want to own their stack. What is the biggest bottleneck you’ve faced when moving your AI agents from prototype to production?