Why multi-agent workflows fail in production

Multi-agent sounds like the obvious answer: parallelize work, specialize agents, go faster. And for s, it works - you can show three agents collaborating on a feature and it looks impressive. In production, the failures are consistent enough that Cognition - the team behind Devin - published a post titled Don't Build Multi-Agents. The GitHub blog ran Multi-agent workflows often fail. Here's how to engineer ones that don't. These aren't fringe complaints. They're structural. Context doesn't travel The foundational problem: each subagent starts fresh.