Building Team Capabilities for AI
How does AI training become everyday practice? Build lasting capabilities through role mapping, team experiments and observable evidence of change.
Participants leave an AI workshop familiar with new tools, encouraged by examples and ready to experiment. Two weeks later, everyday pressure takes over. Adoption depends on a handful of enthusiasts. Capability building becomes more durable when learning sessions connect to the way the team actually works.
In Reskilling in the Age of AI, Jorge Tamayo, Leila Doumi, Sagar Goel, Orsolya Kovács-Ondrejkovic and Raffaella Sadun examine reskilling as a strategic and organizational challenge. Assigning it solely to the HR function narrows that challenge. Source, September–October 2023.
McKinsey’s 2026 management playbook similarly emphasizes the combination of domain knowledge, technology understanding and the ability to lead change. Different roles need to work toward the same business outcome; this does not require everyone to become a developer. Source, pp. 4–5.
Move from a tool list to a role map
A manager makes priority and risk decisions. A process owner sees the complete workflow. A specialist evaluates output quality. A practitioner experiments in daily work and surfaces problems. One person may occupy several of these roles, but their learning needs remain distinct.
For each role, complete this sentence: “After this work, the person should be better able to make the following decision…” It connects the learning objective to an observable behavior.
Make room in the working calendar
The following is a suggested practice to adapt, rather than a universal schedule drawn from research:
- Use a shared opening session to define the problem and quality criteria.
- Let participants select a small experiment from their own work.
- Review unsuccessful outputs alongside successful examples at the next team meeting.
- Keep reusable templates, verification steps and known limitations in a shared place.
The manager’s participation matters. Giving this work time turns learning from an optional extracurricular activity into part of the job. It also creates an opportunity to notice whether existing incentives discourage experimentation.
Add behavioral evidence to completion rates
Attendance and satisfaction are useful measures, but neither alone demonstrates a change in the work. Collect tangible outputs: a stronger decision memo, a verified analysis or a clearer handoff protocol. Reviewing outputs against consistent criteria before and after learning makes progress easier to discuss.
Stratify’s method connects discovery, tailored design, application and measurement. For AI, the starting point remains the work and decisions the team needs to perform better. The appropriate learning agenda follows from that understanding.
