/19

AI Pilot Template

Build a practical 30-day AI pilot plan around one workflow, one owner, one measurable outcome and a real decision at the end.

Start here

10 min

Open +

A useful AI pilot is not “try the tool for a month.” It is a bounded business test. Define the workflow, prove the before state, set the human boundary, measure the same evidence after the test, then keep, change or kill it.

Leave knowing / Leave with a four-week pilot shell you can actually run

The rule

Pilot one workflow. Not “AI.”

The pilot should name one repeated piece of work, one accountable human owner and one business outcome. If the scope is “use AI across the team,” there is no stable before state and no clean decision at the end.

A 30-day window is long enough to see real exceptions and short enough to prevent a vague experiment from turning into permanent shadow infrastructure.

A pilot is a decision mechanism, not a demo with a calendar.

Four weeks

The smallest useful pilot shape.

01
Week 1 / Before

Collect real examples. Record current cycle time, quality, exceptions, inputs, systems, approvals and the human effort required.

Output / A frozen baseline and a bounded workflow

02
Week 2 / Beside

Run the AI-assisted path beside the existing process. Do not remove the human path yet. Log corrections and unsupported outputs.

Output / Evidence about what actually helps

03
Week 3 / Tighten

Narrow the automation boundary to the part that is repeatedly useful. Make verification, permissions and recovery explicit.

Output / A repeatable candidate workflow

04
Week 4 / Decide

Compare against the Week 1 baseline using the same acceptance standard. Keep it, change the boundary or kill it.

Output / A defendable go / change / stop decision

Before day one

A pilot is not ready until these are answerable.

01

What exact workflow is in scope?

02

Who owns the live outcome when the AI is wrong?

03

What is the current baseline?

04

What measurable change would count as useful?

05

How will important outputs be checked?

06

What data and system permissions are required?

07

What would make us stop the pilot early?

08

What decision will we make on day 30?

Try it

Free / Working template

Build the shell of a real AI pilot.

One workflow. One owner. One outcome. Four weeks. No transformation theatre.

Your free output

Your four-week pilot outline will appear here.

The paid Pack goes further: fit, risk, AI/human boundary, controls, build-vs-buy, capacity, evidence and stop conditions.

Next step / AI Pilot Pack

CAD $149 / One time / One real workflow

The free resource gives you a piece. The Pack makes the decision.

Put one real workflow through fit, risk, AI-versus-human boundaries, controls, build-versus-buy, capacity, evidence, stop conditions and a complete 30-day pilot plan.

The workbench is deterministic and your workflow answers stay in this browser. It can recommend that you do not automate the workflow.

Notes / Subscribe

Get the next note.

No schedule. No content calendar. A note goes out only when there is something worth understanding.