The Same Answer
Google, AWS, Microsoft, and Databricks are competing on whether your agents survive production.
Four companies that spend billions trying to be different said the same thing in June. Google shipped an agent platform with cryptographic identity per agent. AWS launched a managed knowledge graph so agents know where to find data. Microsoft reframed the bottleneck as context. Databricks said the visible part of an agent is 1% of the problem.
Every one of them led with operations. Runtime, memory, governance, cost tracking, tracing. The parts that keep agents running after deployment.
This lines up with what Sinch found last week. Seventy-four percent of enterprise agent deployments got reversed after going live. The vendors saw the same data. They are selling the infrastructure to govern agents.
The convergence points to the same problem: what happens after you deploy.