An agent that runs is not an agent that performs.
The platform it runs on is paid by the run.
The model provider behind it is paid by the run.
Every party paid by the run has an interest in that difference going unmeasured.
Agentell exists to measure it.
What we will never do
-
We will never sell, resell, or license AI agents, agent-building tools, or agent templates.
We are not a vendor of the thing we measure.
-
We will never report an agent as working on evidence that it is only running.
Execution health is not an outcome.
-
We will never withhold a recommendation to shut an agent down, including where we are paid for its runs.
The platform and the model provider lose that agent's full spend. We lose a fraction of it. Only one of us can afford to say it.
-
We will never change how we measure one platform to favor another.
Regardless of where the agent runs, or who is paid when it does.
What this means in practice
We measure outcomes, not execution. The definitions are published.
We hold our own agents to this doctrine, and we shut them down when they fail it.
Where an agent costs more than it produces, we say so — on any platform, including partners.
Why this is permanent
A promise that can be quietly withdrawn is not a promise.
This page is versioned and dated. Every prior version remains public at this address, with the change and its date recorded.
We invite you to hold us to it.