Do not wait for perfection. Pilot with three or four users. Give them the daily priority message. Let them complain about what the agent gets wrong. Collect those complaints as the backlog for the next version of the algorithm.
The same approach works for the GTM side. Turn the agent on for one campaign, watch what the routing agent gets wrong, and tighten the rules. Watch a single VP of Engineering get routed to the self-serve flow and use that miss to fix the firmographic enrichment. The pilot is not a test of whether agents work. The pilot is the first iteration of a system that will run in production permanently.
Every piece of feedback is a refinement to the logic. Every edge case is a rule you add to the code. The discipline is to keep the loop tight — pilot Monday, complaints Tuesday, fix Wednesday, ship Thursday, repeat. Not a quarterly review. Not a steering committee. A daily working rhythm where the system gets better because real humans use it on real work and tell you what is wrong.
This is how you operationalize AI in the enterprise. Not with a vendor selection process and a six-month implementation timeline. With a working agent built on inspectable logic, a simple algorithm, a feedback loop, and the willingness to let the system evolve in production.
The teams that wait for the perfect agent ship nothing. The teams that ship a rough agent and tune it for six weeks have something running. The pilot is the product. Treat it that way from the first commit.
— Ry
Sources
Related Essays
Agents in Production: GTM Mesh and the Death of the ERP
The same mesh-of-agents pattern that closes the gap between ad click and revenue is the one that retires the ERP dashboard. Two examples, one architecture.
The Demo Is Not the Deployment
A French CTO with twelve Claude Max seats can ship from his laptop and watch his product team file tickets and wait. That gap is the real problem.
Start With the Pain, Not the Platform
The temptation with agent projects is to start with the technology. That is how you end up with a demo that impresses nobody who actually has to use it.
Key takeaways
- Do not wait for perfection. Pilot with three or four users and let them complain about what the agent gets wrong.
- The complaints are the backlog. Every edge case becomes a rule. Every miss becomes a refinement.
- The pilot is not a test of whether agents work. It is the first iteration of a system that will run in production permanently.
- This is how you operationalize AI in the enterprise — not with a six-month implementation plan but with a working agent and a feedback loop.
FAQ
How do you scope a good first pilot?
Three or four users. One specific workflow they own. A simple, rule-based first version of the agent. Daily output they actually need. A feedback channel where they can complain in real time. Two weeks before the first meaningful iteration. Anything bigger and you are doing implementation theater.
What if the pilot reveals the agent is wrong a lot?
That is the point. Wrong predictions are the most useful signal you can get. Each one is a rule you add or refine. The team that treats the pilot as a feedback engine ships the production system. The team that treats it as a pass/fail test cancels the project at the first miss.