Skip to content

Blog

Double throughput needs a mandate, not another pilot

A pilot asks a team to prove agentic engineering works here. A mandate states the outcome and makes the boring blockers somebody's problem, which puts the uncertainty on the organization.

Written by

Organization and roles

Published: 28 August 2026 Updated: 2 min read

Engineering leaders ask me the same question, almost word for word: “How do we run a good AI pilot?” My answer has stopped being a better pilot design. It is: stop running a pilot.

A pilot is a question, a mandate is a decision

The difference is more than semantics.

A pilot is a question. It asks a team to find out whether agentic engineering works here, on a fraction of the budget, with no promise that anything changes if it does. The team is on trial. The result is uncertain, and everyone treats it that way.

A mandate is a decision. An executive states the outcome: engineering throughput doubles this year. The organization’s job is no longer to find out whether, but to deliver it. The uncertainty moves from “does this work” to “what do we change to make it work.”

That shift is the whole argument. The uncertainty has to land somewhere. A pilot puts it on the technology. A mandate puts it on the organization, and the organization is where it lives.

Pilots succeed and change nothing

I have watched the pilot version fail the same way every time. The pilot succeeds. The team is faster. Then the report goes up, gets a nod, and nothing structural changes, because a pilot was never wired to change anything.

Procurement still takes fifteen approval rounds for a tool license. Developers still pay for agents on personal credit cards while an enterprise token pool sits unused, because no one with authority owned the rollout. The experiment proved the point and the organization absorbed zero of it.

A mandate gives the boring blockers an owner

A mandate is what makes the boring blockers somebody’s problem:

  • Procurement gets a deadline, because the outcome depends on it.
  • The org redesign gets an owner, because doubling throughput needs the work around the line to change.
  • Measurement gets funded, because a mandate has to be checked against real numbers, not vibes.
  • The skeptics get addressed head-on, because you cannot opt out of a mandate the way you can ignore a pilot.

None of that happens for an experiment. All of it happens for a committed outcome.

A mandate still keeps the practices and the gates

This is not a call to be reckless. The mandate names the outcome. It does not skip the practices, the measurement or the gates. It stops pretending the question is whether agentic engineering works, when the real question is whether the organization will change enough to use it.

The constraint is commitment

The constraint was never proof. The technology has been validated, repeatedly, in everyone else’s pilot. The constraint is commitment. A doubling does not happen because a team ran a clever experiment. It happens because someone with authority decided it would, and then changed the organization until it did.

Is your AI effort a question your team is asking, or an outcome your leadership has committed to?

Further reading

  1. When agents write the code, judgment should outrank titles Career ladders reward scope because writing good code at volume used to be scarce. Agents moved the scarcity to judgment, and hiring, leveling, the ladder and pay all need to follow. 2 min readShort post
  2. Coding agents are an organizational-design problem Cheaper local generation pushes the coding-agent constraint into team practice, the shared platform, governance, and learning. One practitioner's four-level model of that shift, held up against what DORA can and cannot support. 6 min read
  3. When engineering speeds up, the bottleneck moves to product A workshop mapped a full SDLC against agents. Almost every step compressed, except deciding what to build next, and most organizations are not staffed for that shift. 1 min readShort post

All 15 posts on Organization and roles

ClientsCase study: A large DIY retailer in Germany

Scaling needs named owners

Coding agents scale across teams when named people own the platform, the team practice, the guardrails, and the measurement. Rollout & scale is where that ownership is built and handed to internal owners, in your stack.

Training on this topic: Agentic Engineering for Engineering and Product Leaders
See Rollout & scale

Want to check one team first? The self-check: twelve yes or no questions, three minutes, instant result, no account. Take the self-check