We have an AI agent fragmentation problem

About this article

Every AI agent works fine on its own — but the moment you try to use more than one, everything falls apart. Different runtimes. Different models. No shared context. No clean way to coordinate them. That fragmentation makes agents way less useful than they could be. So I started building something to run agents in one place where they can actually work together. We have plugins system and already defined some base plugins. The whole architecture is event based. Agents are defined as markdown f...