Execution plan

The first 90 days

Three phases, twelve weeks, and a named owner per week. Written so an internal team can run it without us, and so a board can tell whether it was actually run.

Run it with us

Phase 1: Days 1 to 30, audit
Phase 2: Days 31 to 60, design
Phase 3: Days 61 to 90, execution

Phase 1, days 1 to 30

Audit

Week 1
A single inventory of every active AI pilot, shadow tool, and unofficial use of a public model, across every function. Nothing is shut down; this is an inventory, not a verdict.
Week 2
The three loudest AI-adjacent complaints, each tested for persistence, spread, consequence, and compression, reduced to one or two structural signals written in one sentence each.
Weeks 3 and 4
The organisation scored 1 to 5 on the four readiness pillars, in one room rather than by survey, with one named executive assigned to close the lowest-scoring pillar within sixty days.

Phase 2, days 31 to 60

Design

Week 5
A shortlist built and cut hard through three gates: strategic consequence, measurability, feasibility.
Week 6
A value map for the top one or two candidates, with the baseline number written down before the target.
Weeks 7 and 8
The process mapped end to end, covering trigger, input, human action, proposed AI action, checkpoint, and exception path, with a named person for each of Approve, Override, and Escalate on that specific workflow.

Phase 3, days 61 to 90

Execution

Week 9
A five-sentence pilot charter covering business problem, metric, baseline, quality bar, and stop conditions, signed off by sponsor, operational owner, and risk lead before a single user touches the tool.
Weeks 10 and 11
The pilot run in a bounded setting, with a fully loaded cost comparison built as it goes and the hardest cases deliberately sampled.
Week 12
Evidence brought back to the same room that scored readiness on day 30, answering value, quality, adoption, and risk, and a decision made: expand, stop, or narrow.

The gate at each phase

What has to be showable

Day 30, audit

An inventory of every active pilot and shadow tool, two structural signals in one sentence each, and a four-pillar score with a named owner on the lowest pillar.

"We are exploring several initiatives across the business and building excitement."

Day 60, design

A shortlist of one or two use cases that survived three gates, a completed value map with a written baseline, and Approve, Override, and Escalate named for the specific workflow.

"We have identified a number of high-impact opportunities and are socialising them with stakeholders."

Day 90, execution

A signed pilot charter, results against a pre-written pass line, a fully loaded cost comparison, and a decision: expand, stop, or narrow.

"Early feedback has been very positive and the team is keen to keep going."

Day 30 decision

Which pillar is weakest, and who owns closing it within sixty days. If no name is written down, the programme has not started.

Day 60 decision

Which one or two use cases survived the three gates, and who holds Approve, Override, and Escalate on the specific workflow.

Day 90 decision

Expand, stop, or narrow, argued against the pass line written on day 61 and a fully loaded cost comparison.

Next step

Run the ninety days with a senior team beside you

The AI Readiness Diagnostic covers phase one and two. The Prototype Sprint covers phase three where a build is the right answer.

AI Readiness Diagnostic Prototype Sprint