Free · The Health Pack

The AI Programme Health Pack

Two tools for whoever is answerable for AI at work. A six-question Health Check that shows where your programme is strong and where it has a gap, and the scorecard for the AI Champions who run it.

1 The Health Check Six questions about your programme, a score out of 12, your strongest area and your biggest gap. About three minutes.
2 The AI Champion scorecard Seven competencies, from problem diagnosis to risk triage, against the standard a Champion is expected to meet. About ten minutes a person.

2 · The AI Champion scorecard

Seven competencies, against the Champion standard

An AI Champion is a trained person inside the business who helps colleagues identify, scope and prototype practical AI opportunities, and knows when an idea needs specialist technical, security, legal or data review. Not an AI engineer.

1: Awareness
2: Practitioner
3: AI Champion
4: Advanced Champion

A Champion is expected to reach 3 in every competency, except Automation and integration, where 2 is enough. Score what the person has actually done, not what they could probably do.

Who are you scoring? (optional)

1 of 7 Problem diagnosis

What good looks like. Turns a vague idea into a problem statement with an owner, impact, scope and a measurable baseline, and challenges whether AI is needed at all.

1 Awareness Describes an idea in terms of the tool someone wants, rather than the problem it would solve. 2 Practitioner Writes a clear problem statement for their own work, with who is affected and how often it happens. 3 AI Champion Helps a colleague turn a vague idea into a problem statement with an owner, impact, scope and a baseline to measure, and asks whether AI is needed. Expected for a Champion 4 Advanced Champion Spots when the stated problem is a symptom of another one, reframes it with the people involved, and is comfortable concluding that AI is not the answer.

2 of 7 Workflow design

What good looks like. Maps the current and proposed workflow, places human checkpoints, states the expected improvement, and removes unnecessary work before automating.

1 Awareness Thinks of AI as replacing a whole task rather than improving one step in a process. 2 Practitioner Maps their own process step by step and points to where AI could help. 3 AI Champion Maps the current and proposed workflow with a colleague, places human checkpoints, states the expected improvement and removes unnecessary steps first. Expected for a Champion 4 Advanced Champion Redesigns workflows across teams and hand-offs, with checkpoints in proportion to the risk of each step.

3 of 7 Prompting and AI interaction

What good looks like. Builds reusable, testable interactions with clear inputs, instructions, constraints and outputs.

1 Awareness Uses chat tools with short, one-off requests and accepts the first answer. 2 Practitioner Writes structured prompts with a role, context, constraints and the format wanted, and gives specific feedback when the output is wrong. 3 AI Champion Builds reusable instructions or assistants other people can use, with defined inputs and outputs, tested on realistic examples. Expected for a Champion 4 Advanced Champion Designs instructions that hold up across difficult and unusual cases, keeps versions, and teaches others to work out why an output went wrong.

4 of 7 No-code prototyping

What good looks like. Builds a working, testable MVP without Engineering building the first version, and knows when it has passed the safe boundary.

1 Awareness Knows no-code tools exist but has not built with one. 2 Practitioner Follows a guide to build a simple form, assistant or flow for their own use. 3 AI Champion Builds a working MVP a small group can test, without Engineering, and knows when it has gone beyond what is safe to run that way. Expected for a Champion 4 Advanced Champion Builds MVPs for several teams, reuses what worked, and hands over to Engineering with a clear record of what was built and why.

5 of 7 Evaluation and measurement

What good looks like. Works through baseline, hypothesis, test, evidence and decision, knows that time released is not money saved, and recommends Scale, Change or Stop.

1 Awareness Judges an AI output by how well it reads. 2 Practitioner Checks outputs against a few examples and records how long a task took before and after. 3 AI Champion Sets a baseline and a hypothesis, tests on at least ten realistic cases with a test log, and recommends Scale, Change or Stop from the evidence. Expected for a Champion 4 Advanced Champion Designs measures that stand up to a finance director, looks for evidence against their own idea, and is as comfortable recommending Stop as Scale.

6 of 7 Automation and integration

What good looks like. Automates straightforward AI-enabled workflows safely with tools such as n8n, Make, Zapier or Power Automate, with approval points and basic error handling.

1 Awareness Knows what an automation tool does but has not built a flow. 2 Practitioner Builds a simple flow from a trigger to an action for their own work. Expected for a Champion 3 AI Champion Automates a straightforward AI-enabled workflow with approval points and basic error handling. 4 Advanced Champion Connects forms, databases and business systems safely, handles failures and permissions, and knows where a production integration needs Engineering.

7 of 7 Risk triage and escalation

What good looks like. Triages data, privacy, security, accuracy, fairness, human impact, legal, dependency and vendor risk, then proceeds as a low-risk MVP, adds controls first, or escalates to the right people.

1 Awareness Knows some data should not go into AI tools but is unsure where the lines are. 2 Practitioner Follows the company AI policy and recognises personal and sensitive data in their own work. 3 AI Champion Triages an idea across data, privacy, security, accuracy, fairness, human impact, legal and vendor risk, then proceeds, adds controls or escalates to the right person. Expected for a Champion 4 Advanced Champion Anticipates risks others miss, writes risk notes specialists can act on, and helps shape the company rules.

0 of 7 scored

The profile

A rating is a starting point for a development conversation, not evidence. The real test is taking one real idea from a problem statement to a Scale, Change or Stop recommendation.

Competency	1	2	3	4
Problem diagnosis
Workflow design
Prompting and AI interaction
No-code prototyping
Evaluation and measurement
Automation and integration
Risk triage and escalation

Filled squares show the score. The outlined square marks what a Champion is expected to reach.

The AI Programme Health Pack

Open the pack

Six questions about your AI programme

Book a 30-minute AI Programme Review

Seven competencies, against the Champion standard