The Team AI Academy: Upskilling That Changes Real Work

A useful baseline starts with work evidence. Confidence surveys reveal how people feel. Course completions reveal exposure. Neither shows whether someone can apply AI safely and repeatedly inside a real workflow.

Choose one recurring task for each participant or role group. Review a recent work product and ask how the task was completed. Score four capabilities: framing the task, selecting useful context, evaluating the output, and integrating the result into the workflow. Use one shared scale from zero to three: no evidence, attempted, repeatable, and repeatable with verified quality.

The evidence matters more than the number. A low score should point to a visible constraint, such as weak instructions, missing source material, shallow review, or an isolated experiment that never reaches the final work product. That constraint becomes the skill gap the academy must address.

Keep the scorecard narrow. It is a curriculum decision tool, not a verdict on someone’s value or future potential. Record where evidence is unavailable and name the assumption that needs validation during the first practice session.

The baseline scorecard showed where fluency breaks inside real work. Use that evidence to build a track for one role and one recurring workflow. A shared catalogue may support the programme, but it cannot define the path. Different roles face different decisions, source material, risks, and standards of quality.

Design backwards from the changed work product due on day 30. Name what the participant should produce, then identify the smallest capabilities needed to produce it well. Each capability becomes a skill in the track. Each skill must immediately connect to a live task from the selected workflow and leave behind an observable output.

Sequence by dependency. Start with the capability required for the next one to work. Keep the track narrow enough to practise during the 30-day cohort. Extra topics create motion without transfer.

A useful track has a clean chain: baseline gap, required skill, live practice task, observable output, changed work product. That chain makes curriculum decisions visible. It also gives the weekly ritual something concrete to run and the later scoreboard something meaningful to measure.

A role-based track only changes work when practice enters the working week. The weekly session is the delivery mechanism: a protected moment where participants apply one target skill to a task already on their desk.

Keep instruction brief. Most of the session belongs to doing, reviewing, and committing. Participants arrive with a live task from the workflow selected in their baseline scorecard. They use the week’s skill to change that task, then capture evidence before leaving. Evidence could be a revised work product, a tested prompt with its result, or a documented decision that improved the workflow.

Peer review should judge usefulness, not performance. Ask whether the application made the output clearer, faster, safer, or easier to reuse. Choose the criterion that matches the diagnosed gap. When the work falls short, the participant records what to adjust next time.

Close with a specific next-use commitment tied to the same recurring workflow. Name the task and deadline. This turns a workshop into a cadence. Across four weeks, the cohort creates a visible trail of application that the next chapter can measure. Where quality or time improvements are claimed, treat them as assumptions until the team compares real before-and-after evidence.

A full session can still leave the workflow untouched. Your scoreboard must reveal whether people attended, applied the skill, and changed a real work product. These are separate signals.

Participation shows reach. Record active cohort members and attendance so you can spot access or scheduling problems. Application shows transfer. Count a participant only when they provide evidence that AI was used in the recurring workflow selected during the baseline. A verbal claim is too weak. Use a link, screenshot, prompt record, process note, or reviewed output.

Changed work products show the programme’s result. By day 30, compare a participant’s earlier output with a reviewed version produced through the track. Describe the observed change in work terms such as time taken, clarity, accuracy, usability, or risk. Where no reliable comparison exists, label the claim as an assumption to validate.

Keep the scoreboard visible to the cohort and update it after each weekly ritual. Use counts and rates together. A percentage can flatter a tiny cohort; a count exposes the base. Avoid ranking individuals. The purpose is programme judgment: scale a track with repeated application and credible work change, revise one with participation but weak transfer, and stop one that produces no useful evidence.