Every enterprise conversation about agentic AI eventually arrives at the same question: Is the model good enough? It’s the wrong question. Model performance has advanced faster than many experts have predicted. What hasn’t been solved—what’s actually holding companies back from letting agents act—is trust: trust in the data an agent is reasoning over, trust in the context it’s using to make a call, and trust that its reasoning can be inspected and, if necessary, reversed. Coagency: A New Kind of Decision-Maker
In Agentic Intelligence, we introduce the idea of “coagency,” which is the recognition that once you deploy an agent that can act, not just recommend, you’ve created a new kind of joint decision-maker. The agent isn’t a tool waiting for a human to pull the trigger. It’s a participant. And participants in any relationship, human or otherwise, earn authority through demonstrated reliability, not through a spec sheet.

Full commentary : Manish Sood and Venkat Venkatraman argue that the adoption of agentic AI hinges not on model performance, but on building trust.