WonderLead · Page 1 of 4
The Human-First AI Adoption Checklist
Ten moves. Four weeks. One method.
This is the method we run with teams, written down so you can try it yourself first. It comes from a bottom-up AI adoption programme at AWS Berlin: nine months, seven pizza teams, then three organisations sharing a floor — Party Rock, AWS Console Platform, AWS UX. Different products, same problem. Nothing in it is a tool decision.
If it works on its own, use it and don’t book anything. If it shows you a gap you don’t want to close alone, that’s what a conversation is for.
The order is the method
Most enablement fails because it leads with technology and skips the humans. People are hesitant, some go quiet, others opt out, and a dashboard tracking their usage only makes it worse. These ten moves reverse that sequence: safety before skill, every role included, real work as the material, then — only then — let it spread.
How to use this
- Print the one-pager. Put it next to your 1:1 notes.
- Work through one week at a time. Do not skip Week 1 because it feels “soft.”
- Measure behaviour first (who shows up, who asks, who demos). Outcomes come after.
WonderLead · Page 2 of 4 · Weeks 1–2
Safety first, then meet people where they are
01Make it voluntary.No mandates. No appointed “AI champions.” Compulsory learning produces attendance, not practice. At AWS, neighbouring teams joined because they wanted in, not because they were told to.
02Take usage tracking off the table.People will not experiment if they think they are being scored. Phase one has no dashboard and no seat-licence report. You are buying psychological safety, not a metric.
03Put the sessions on work time.If it is extra-curricular, only the already-converted show up. Lunch-and-learns after a full day are a tax on the curious. Put it on the calendar as work.
04Lead with a failure of your own.The first attempt on this programme led with a multi-agent scrum system and got resistance. Leading with the technology is the same mistake everyone above you is making. Say that out loud.
05Ask, don’t prescribe.Fifteen-minute 1:1s: how are you using AI now, and what would make it safer to try? You will be surprised. Meet people at their actual starting point, not the one in the vendor deck.
06Include every role from day one.AI is not a developer tool. Non-technical people often need the most support and bring the most useful insight. Role-specific work, not a generic prompt workshop.
07Use real work as the material.No sandbox. No fictional exercises. On-call agents, design-system automation, the ticket that is already on the board. The habit has to survive past the novelty.
WonderLead · Page 3 of 4 · Weeks 3–4
Learn in public, then let it spread
08Run a weekly sharing session.Failures are the content. Half-finished demos count. Honest review replaces polished theatre. This is how a series of demo sessions became a community that ran for more than six months.
09Let learners become the teachers.The room changes the moment someone who had never touched AI demos an agent they built. Peer teaching is the distribution mechanism. You are not the hero of this story; you are the person who made it safe.
10Don’t mandate the neighbouring teams.Make it easy to join, and let them opt in. One team to seven. Seven to three organisations on one floor. A cross-org guardrails working group formed because people wanted one, not because a slide said they should.
What you should start seeing
- People choosing to participate, and asking more questions.
- Different roles collaborating on the same experiments.
- Anxiety down and real use up — in the actual work, not in a sandbox.
What you should not do yet
Do not roll out a usage dashboard. Do not appoint champions. Do not buy another tool to “fix adoption.” If the ten moves are not sticking, the gap is leadership, not licences.
WonderLead · Page 4 of 4 · Measure, then decide
Behaviour first. Outcomes after.
At AWS Berlin we did not measure adoption as seats used. We measured whether people showed up, asked, demoed, and taught. Outcomes (agents in production, a guardrails group, three orgs on one method) arrived after the behaviour changed.
A simple scorecard for the four weeks
- Voluntary attendance — are people choosing to come back?
- Questions asked in the open — is it becoming safer to be a beginner?
- Cross-role experiments — is this still “an engineering thing”?
- Peer demos — have learners started teaching?
- Opt-ins from next door — is it spreading without a mandate?
If you’d rather know where you stand first
The AI Adoption Assessment scores your team across Daily Practice, Safety to Learn, and Leadership System. No login. The score is shown before anything is asked of you. site.wonderlead.tech/companies/ai-adoption/
If you’d rather not run it alone
WonderLead runs this as a facilitated cohort with your teams, on work time, using your real work. The AI-Native Team Programme. If you’d rather not run it alone, book a conversation and we’ll talk it through.
How AI adoption spread across three organisations at AWS Berlin: site.wonderlead.tech/companies/aws-berlin/
Adoption is a leadership test. The tools are the easy part.