AI Champions
The format
- 6weeks, each with a live workshop and a hands-on build clinic
- 8hours a week, about half of it live with a coach
- 48hours in total, spent on a real problem from the champion's own team
A few people per team, each with their manager's backing.
4.8 / 5 from 78 Google reviews
1,000+ learners trained
Trusted by teams at
Who it is for
The people colleagues already turn to
Champions are not chosen for job title. They are the people close to the work who will carry the change, and they need domain depth plus enough AI fluency to translate between the two. Nobody has to become an engineer.
A domain expert first
They know their function inside out and keep looking for a better way to run it.
Someone colleagues listen to
That trust is what lets a new tool get adopted, not just demoed.
Naturally curious
They already have the basics and want to go further on a problem their team keeps hitting.
A multiplier, not a one-off builder
The programme is designed to cascade: what one champion learns reaches their team.
Each champion comes ready with
- A team problem their manager has endorsed.
- Access to the AI tool their team already uses, and team data they can already use.
- Foundation-level AI fluency (prompting, connectors and projects), which we build on rather than re-teach. If they do not have it yet, start with AI Foundations.
The format
Six weeks, about eight hours a week
Around 48 hours in total, fitted around the working week. Each week has three parts, which we call Acquire (learn the idea), Apply (use it live with a coach) and Accelerate (build with it on the team's own problem).
A few people per team
Each team puts forward a few champions, and they learn together in a small cohort.
Manager-endorsed
The manager backs the place and the problem. In week 1 the champion agrees the use case with them; in week 6 they bring back a proposal the manager and function head can approve.
Acquire, short pre-work
Done in the champion's own time before each week's live sessions.
Apply, about 4 hours
A two-hour workshop plus a two-hour hands-on build clinic, in person, online or hybrid, with a practitioner coach. Coaches are confirmed per cohort.
Accelerate, about 4 hours
Building the team's own tool between sessions, so the work pays back while they learn.
In your own AI tool
Champions build in the AI tool their team already has access to, so what they make is ready for the team to use.
Curriculum
From a team problem to a pilot that is proved to work
Two skills come first, taught in week 1 before any product opens. Every week after that builds the same pilot, on the champion's own use case.
Skill first
Identifying use cases
Look at your own function and spot where AI genuinely helps, and where it does not. Prioritise on value and feasibility, and say no to the tasks that are not a fit.
Skill first
Process architecture
You cannot automate what you cannot describe. Break the workflow into steps, decisions and exceptions, and name the inputs, the data and who owns each handoff.
Week 1
Find the use case, and map it
Where AI genuinely helps in your function, when a task needs an agent and when a simple model will do, and how to put a number on the opportunity before anything is built. You define what a good result looks like and how you will measure it, and agree the problem with your manager. Then you diagram the process as it is and as it should be, with no product open yet.
You leave the week with: A sized use case and a process diagram
Week 2
Turn the diagram into a working pilot
Structuring prompts for reliable, repeatable output, and giving the model the right documents and data so it performs at team scale. Connectors, projects and reusable templates in practice. You build on practice data shaped like your own use case, and you build it modular, so real data swaps in later without a rebuild.
You leave the week with: A modular pilot, ready to swap in real data
Week 3
Automate it, and keep a person in control
Where workflow and non-workflow automation tools each fit. Reusable skills, projects and connectors for repeatable automation you can build without engineering. Why an output will not settle, and how to fix it: restructure the context, tighten the constraints, or break the task into steps. Then you design the checkpoints where a person stays in control.
You leave the week with: An automation with a human checkpoint
Week 4
Prove it works before you trust it
Why and how you measure a pilot before you rely on it. You write the ground-truth answers your evaluation is measured against: test cases, pass and fail, edge cases and regression checks. You learn when human review is the right check, when a model can judge, and when you need both, and you apply guardrails and safe data handling.
You leave the week with: An evaluation harness mixing automated and human review
Week 5
Run it live, and measure what it saves
A facilitated hackathon with invited domain experts from your organisation, where the pilot goes in front of real users. Then the three numbers that decide whether it scales: hours reclaimed, volume handled without adding headcount, and whether quality held against the manual process it replaces.
You leave the week with: Proof it saves time and holds quality
Week 6
Plan how AI scales across your function
A roadmap for AI in your own function: which use cases come next, in what order, and why. For each one, you decide whether you can build it yourself or it needs engineering support. You package your tool so a teammate can pick it up, and turn the case you pitched at the hackathon into a proposal your manager and function head can approve.
You leave the week with: A prioritised AI roadmap for your function
Afterwards
What a champion can do at the end of six weeks
Scope and justify a use case. Pick a team-level problem, size its value, and win their manager's backing before anything is built.
Make the model do the job. Give it the right context at team scale, and steer and debug it when it will not do the thing.
Prove it before trusting it. Build an evaluation harness and apply guardrails, so the team relies on results that have been checked.
Hand it on. Package the tool as a reusable team asset (a skill, a project or a template), pilot it with the team, and pass the skills to colleagues.
Build the next one. Leave with a pilot proved to work, the skill to build the next one, and a plan for where AI goes next in their function.
How we measure it
Measured on the work, not on attendance
Before and after
A short assessment in week 1, and the identical assessment in week 6.
On the pilot itself
In week 5: hours reclaimed, volume handled without adding headcount, and whether quality held against the manual process it replaces.
Where it leads
Our funded Level 4 programmes take champions further
Every champion who completes the six weeks receives an iO-Sphere certificate of completion. For the ones who want to go deeper, our government-funded Level 4 programmes run alongside the job.
Funded Level 4
AI Transformation
Built on the national Level 4 Business Analyst standard (ST0117). About 15 months of training plus a three-month final assessment, at around six hours a week: two hours of live coaching, two of self-paced learning and two of coach-supported project work in the champion's own team.
It goes deeper on the same ground: where AI fits the organisation and where it does not, building an AI agent for a real task, business cases, and leading change from pilot to strategy.
About AI Transformation →For employers
Three Level 4 programmes, government-funded
AI Transformation, Advanced Data & AI and Data & AI Governance are our three Level 4 programmes. All three are endorsed by The AI Board, an Ofqual-regulated awarding organisation.
They are funded through the Growth & Skills Levy. What an employer contributes depends on the learner's age and your levy position, and we guide your team through the setup.
How funding works for employers →Rolling AI out across a whole function? The whole-organisation programme plans your champions and the rest of their colleagues together.
Questions
AI Champions FAQs
Who should we put forward as champions?
+
The people colleagues already turn to: domain experts who know the work inside out, are curious about better ways to run it, and will pass what they learn to their team. They do not need to be managers. Each one comes with their manager's backing and a team problem their manager wants solved.
How much time does it take?
+
About eight hours a week for six weeks, around 48 hours in total. Roughly four hours a week are live: a two-hour workshop and a two-hour build clinic with their coach. The rest is short pre-work and building their team's tool between sessions.
Do champions need to be technical?
+
No. They need deep knowledge of their own function and enough AI fluency to translate between the two. The automation they build needs no engineering, and their week 6 roadmap marks which next steps would need engineering support.
Do they build on our live data?
+
They start on practice data shaped like their own use case, and build the pilot modular so your real data swaps in later without a rebuild. Guardrails and safe data handling are taught in week 4, before anyone relies on the output.
How is it priced?
+
We price team training per engagement, once we know who is taking part and what they are working on. The government-funded Level 4 programmes are funded separately, under the government's own funding rules.
Talk to us about your champions
Tell us about your teams and the problems they keep hitting. Expect a reply within one working day.