Catalogue › Champions › champion-learning

Champion learning

Designs the learning offer behind the Champion journey - training-route sessions and guided exercises for each capability, exemplars and clinics for the evidence route, Challenge preparation, refreshers when Copilot changes and advanced learning for recognised Champions - shared with the workforce where possible. Use when capabilities need a training route, when sessions feel too basic, or when material is stale.

Department

Champions

Lifecycle

Enable, Scale, Improve

Maturity

explorer, adopter, scaler, ai-enterprise

Risk tiers

0, 1

Platforms

copilot-chat, m365-copilot, copilot-cowork, m365-agents

Editions

starter, scale, enterprise

Reviewers

none

Status

draft

Copilot Cowork

Champions plugin for Copilot Cowork, or this skill on its own: Download SKILL.md

Licence

CC BY 4.0, by Simon Foote

Purpose

The capability framework says what a Champion must be able to show; this skill designs the learning that helps them get there and keeps them current afterwards. It builds the training route for each capability, supports people taking the evidence route, prepares Champions for their Challenge, refreshes practice when Copilot changes and offers advanced learning to recognised Champions. It usually goes wrong when learning becomes the gate rather than one route, when sessions are pitched at beginners and lose experienced people, when material describes last quarter's product, or when Champion learning is run as a closed club that duplicates the workforce learning calendar.

When to use

• The capability framework is agreed and each capability needs a training route people can actually take.
• Experienced Champions find sessions too basic, or newer ones find them too fast.
• People choosing the evidence route ask what good evidence looks like and get no help.
• Copilot has changed and the Champion material no longer matches what people see.
• Recognised Champions ask for something deeper, and the programme has nothing to offer them.

When not to use

• Defining the capabilities, routes and verification criteria themselves - use champions:capability-framework.
• Designing progressive learning pathways for the whole workforce - use learning-and-skills:ai-learning-pathway.
• Advanced pathways for makers and specialists beyond the Champion programme - use learning-and-skills:advanced-pathways.
• Building safe practice environments and sandboxes - use learning-and-skills:practice-labs.
• Writing detailed facilitator guides for individual sessions - use learning-and-skills:facilitator-guide.
• Community rituals such as calls and prompt clinics - use champions:champion-community-design.

Questions to ask first

1. What do the capability framework's objectives and verification criteria require, capability by capability?
2. What learning does the workforce calendar already offer, and which sessions could count towards the journey?
3. Who will deliver Champion learning - the learning team, the adoption team, recognised Champions, a partner - and with what time?
4. Is there a safe practice environment, and is agent building approved for Champions at all?
5. How will material be kept current when Copilot changes?

Required inputs

• The capability framework, with objectives, training and evidence options and verification criteria, from champions:capability-framework.
• The workforce learning calendar from copilot-adoption:launch-plan or learning-and-skills:ai-learning-pathway.
• The Challenge options and criteria from champions:champion-challenges.
• The organisation's position on agent building for Champions, and any practice environment from learning-and-skills:practice-labs.
• Delivery capacity: who can teach, how often, and in which formats.

If agent building is not approved for Champions, design the agent capability learning around understanding and judging use cases, and record the configuration element as deferred. If no safe practice environment exists, use fictional material on approved tools and record the gap.

Decision framework

Map learning components to the journey. Each training route ends in the same evidence the evidence route requires.

Decide what is open and what is Champion-only.

Practitioner approach

Start from the verification criteria and work backwards. For each capability, ask what someone must be able to do to produce evidence that meets the criteria, and design the shortest learning that gets them there. If a session does not help people meet a criterion, it belongs in the wider workforce calendar, not the Champion route.

Keep training one route, not the gate. Experienced colleagues should be able to skip straight to evidence. Support them with exemplars of strong and weak evidence, written with fictional material, and short evidence clinics where they can test a draft submission. This respects their time and raises the quality of submissions, which helps assessors.

Make every training route practical. Each capability's training ends with a guided exercise on the participant's own work, producing the evidence the criteria ask for. People learn to verify Copilot output by verifying it, not by hearing about it. Use realistic fictional material in group sessions and remind people never to bring personal or confidential data into shared exercises.

Share foundation learning with the whole workforce. Sessions on Copilot Chat and the Microsoft 365 apps should be open to everyone and recorded, with attendance counting towards the Champion journey. This keeps the programme from becoming a closed club and means the same investment serves both foundation adoption and the network.

Treat agent capability carefully. Learning covers what agents are, how to judge a good use case, ownership and governance, and testing basics. Where the organisation approves Champions to build, add a guided build in an approved environment, using Agent Builder or the tool governance allows. Where it does not, the capability is about judgement and routing, and the learning says so honestly.

Prepare people for the Challenge. A short scoping workshop on picking a real team problem, checking safety and scope, and leaving something reusable, followed by mentoring from a recognised Champion, does more than any further session.

Keep material current. Copilot changes often, so give each module an owner and a review point, and run a short refresher whenever a change alters what people see or what the guidance says. Check current product behaviour against the source records rather than relying on last quarter's slides.

Finally, give recognised Champions somewhere to grow: advanced sessions, early sight of new capabilities in governed pilots and routes into specialist pathways, each with an owner who has agreed to run it.

Outputs

• A Champion learning map showing, for each journey stage, the training route, evidence-route support, delivery owner and rhythm, recorded with the champion-capability-framework template.
• Guided exercises and exemplars for each capability, using fictional material.
• A learning calendar marking which workforce sessions count towards the journey, aligned with the adoption-plan learning calendar.
• A content maintenance plan with module owners and review points.

Microsoft platform considerations

Learning must reflect the Copilot experiences Champions actually have: Copilot Chat, Copilot in the Microsoft 365 apps, Copilot Cowork and, where approved, agent building in Agent Builder or Copilot Studio. Capabilities, naming and availability change, so check references/sources.md before each refresh and point Champions to Microsoft's current learning resources rather than reproducing them.

Adoption considerations

Champion learning is part of the adoption system. Opening foundation sessions to the workforce and aligning refreshers with the reinforcement cycle means Champions arrive at their teams with the same current guidance the programme is promoting.

Knowledge considerations

Learning materials, exemplars and recordings are knowledge assets: give each an owner, a version and a review date, and retire superseded material so Champions do not learn outdated practice.

Security and privacy considerations

Exercises use fictional or approved material. Participants are reminded not to bring personal or confidential data into group sessions or evidence. Agent building happens only in approved environments with the organisation's controls. Recordings follow the organisation's retention rules.

Responsible AI considerations

Every module teaches verification, accountability and the limits of AI output, and the agent capability covers when an agent should not be built. Learning never rewards volume of use.

Evaluation requirements

Not applicable as solution evaluation, because this skill designs learning rather than an AI solution. Learning is reviewed by whether participants on the training route go on to meet the verification criteria, at programme level.

Value measures

Owned by the Champions programme owner with the learning lead, with baselines from the first cohort: progression from training route to verified capability, evidence first-time pass rates at programme level, time to complete each stage, and attendance from the wider workforce at open sessions. Reported at programme level, never as individual rankings.

Scale notes

Starter

Foundation sessions shared with the workforce, one guided exercise and exemplar per capability, and a refresher when Copilot changes, delivered by the programme owner or a partner.

Scale

A full learning map with evidence clinics, Challenge workshops, mentoring and module owners, delivered with the learning team and recognised Champions.

Enterprise

A learning offer shared across regions with localised delivery, advanced sessions and pilots through the CoE, and specialist pathways connected to learning-and-skills:advanced-pathways.

Hand-offs

• To champions:capability-framework when learning shows the criteria are unclear or unrealistic.
• To learning-and-skills:facilitator-guide to write guides for each session.
• To learning-and-skills:practice-labs where a safe practice environment is needed.
• To champions:recognition-and-progression for advanced access and specialist pathways.

Reviewer gates

Not applicable as a formal gate, because learning design does not release an AI solution. Guided agent builds use only environments and tools already approved under the organisation's governance.

Failure modes

• Training becomes the gate, and experienced colleagues leave rather than sit through beginner sessions.
• Sessions describe features rather than producing evidence, so participants cannot meet the criteria afterwards.
• Material lags behind product change, and Champions teach outdated practice.
• Champion learning duplicates the workforce calendar and becomes a closed club.
• Agent capability promises building that governance does not allow.

Sources

references/sources.md in this skill lists the source records behind this skill, what each is authoritative for, how volatile it is and what you may do with it. Product availability, licensing, limits and preview status change often, so check the current source before stating them, and cite what you used.

Never

• Make training the only route to verified capability.
• Use personal or confidential data in group exercises or exemplars.
• Teach agent building in environments the organisation has not approved.
• Measure learning by attendance alone or rank participants.