**Our Take: The Real Agent Gap Isn't Technology, It's Honesty**
Enterprises are making a bold bet on agentic orchestration, and the direction is clear. They are consolidating onto model-provider platforms at a stunning pace, with Anthropic's Claude leading the charge by a wide margin. The logic is sound: choose the orchestration layer closest to the model you trust most, and measure success by reliable, multi-step execution. That is the right ambition. But the latest Pulse Research tells a more grounded story. Most organizations are building the highway before they have the cars on it. When asked to assess their own portfolios honestly, 71% admit that a quarter or fewer of their deployed "agents" are true multi-step workflows. Most are still single-prompt chatbot wrappers. That is not a failure of effort; it is a gap between intent and execution that defines this moment.
The gap is not just about maturity; it is about architecture and control. As enterprises standardize on provider platforms, they are simultaneously moving to insulate themselves from those same providers. The expectation of a hybrid control plane, provider-native plus external orchestration, is now the dominant posture, with 51% expecting this structure by the end of 2026. Vendor lock-in has overtaken security as the primary fear, and it is easy to see why. When the platform is the model, the model is the moat. Enterprises are responding by planning to build in-house control logic, standardize on a single framework, and push agents from sandbox to production. The intent to change is nearly universal, with 68% planning to adopt a new, additional, or replacement platform within the year. The most concentrated layer of the agentic stack is also the least settled.
Here is what concerns us most: the fiscal control gap. More than a quarter of enterprises have no real-time, programmatic way to stop a runaway agent before the bill arrives. They learn of the damage from the logs, after the fact. That is not a budgeting problem; it is a governance problem. If you cannot stop an agent, you cannot trust an agent. And trust is the only thing that moves this from experimentation to production. The technology is not the bottleneck. The discipline to deploy it responsibly is. Enterprises are not failing because the platforms are weak; they are failing because they are treating chatbot wrappers as if they were orchestrated systems. That is the trap, and it is a choice.
The path forward is not to abandon the platforms or to double down on a single vendor. It is to demand more from the control plane you own. The hybrid model is the right instinct, but only if it comes with real-time cost controls and a willingness to measure success by reliability, not hype. The enterprises that will lead this wave are not the ones with the most sophisticated models; they are the ones with the most honest assessments of what their agents actually do. The ambition is correct. The architecture is emerging. The missing piece is the discipline to close the gap between what we call an agent and what we deploy. The tools are ready. The question is whether your organization is.