AI adoption sticks when you start with one repeated, low-risk task, build the AI step into a workflow people already use, keep a person deciding, measure and share the result, and have leaders go first. Those are the five conditions I plan against. Pick that task first, then track breadth, depth and momentum and review after 90 days.
By Johannes Sundlo, AI & Future of Work Advisor. I help leaders and HR teams turn AI into adoption that sticks, through keynotes, workshops and change programs.
AI adoption means putting AI to use in everyday work and checking whether it helps. This guide covers the tasks to start with, the support people need and the measures to follow when moving beyond a pilot.
What “AI adoption” means
AI adoption is a sustained change in how people do their work with AI. Measure its use in specific tasks, the quality of the output and the time spent checking it. Licence purchases and pilot launches alone do not show whether work has improved.
What can slow AI adoption?
- It’s bolted on, not built in. AI lives in a separate tool people forget to open instead of inside the workflow they already use.
- No early, visible win. A 12-month transformation with nothing to show after month one loses people’s attention.
- Fear goes unaddressed. People quietly worry that AI replaces them, so they keep their distance.
- It’s treated as IT’s job. Teams need support with both the tools and the changes to their work.
The “Adoption that Sticks” model
I use the following five conditions when planning AI adoption with leaders and teams. They are a working framework to test against your situation, rather than a guarantee of results.
Two studies describe different aspects of workplace AI. Gallup reported in June 2025 that 40% of U.S. employees used AI in their role at least a few times a year. In a separate international study published in September 2023, Randstad found that 13% of more than 7,000 respondents had been offered AI training in the previous year. The studies cover different populations, dates and questions; they do not measure one shared adoption gap.
- A painful, specific starting point. Begin with one repeated, low-risk task people already dislike, not a strategy.
- Built into the workflow. The AI step lives in the template, checklist, or doc people already use.
- Human in the loop. AI drafts, a person decides, so people trust the output and own the result.
- Visible local proof. Measure the time and quality before and after a change, then share the result with the team.
- Leaders who go first. When leaders use AI out loud, everyone else gets permission to.
Where to start, by role
Where you start depends on your role. These guides go into the detail for each:
- HR teams: see AI for HR: a practical guide to adoption that sticks.
- Leaders and executives: see AI for Leaders: a practical guide to driving adoption.
- Recruiting and talent: see AI in Recruiting: a practical guide.
How to measure AI adoption
I track three signals rather than licence counts: breadth (how many teams use AI in real work without being told to), depth (is it touching meaningful tasks, not toys), and momentum (are new use cases appearing on their own?). When all three rise together, AI has left the pilot and become everyday work.
AI change management: a five-phase framework
What is AI change management? AI change management means helping people use AI in their work, with clear responsibilities and follow-up. I use five phases: assess current use and concerns; let leaders practise; train teams on selected tasks; agree rules for tools, data and review; and measure use, quality and time. The table below is an example plan to adapt to your organisation. My 2025 State of AI in HR survey found that 81% of those answering the relevant question reported employer-provided generative AI and 33% reported formal training. These are separate measures, not evidence that a particular training approach causes adoption.
| Phase | What happens | Owner | Signal it worked |
|---|---|---|---|
| 1. Readiness | Inventory of current AI use, three painful tasks per team, a short fear survey | HR with one executive sponsor | A shortlist of tasks people want help with |
| 2. Leaders first | Executives use AI on their own week and show it, for four weeks | CEO and leadership team | Every leader can name what they used AI for last week |
| 3. Role-based enablement | One hour of foundations, one real task per team, prompts saved and named | Team leads with HR | Each team has three named prompts in use |
| 4. Guardrails | One-page policy: data that never goes in, who signs off, approved tools | HR, IT and legal | People know which tools and data are approved and who can answer questions |
| 5. Measurement | Breadth, depth and momentum reviewed every quarter | Executive sponsor | New use cases appear without being asked for |
How this differs from ADKAR
Prosci’s ADKAR model covers Awareness, Desire, Knowledge, Ability and Reinforcement. In my approach, leaders practise with the tools early so they can discuss their capabilities and limits with their teams. The sequence should reflect what people already know and what is preventing use. I work through these questions in executive AI workshops and change programs .
Common questions about AI adoption
What can make an AI pilot stall?
A pilot can stall when nobody owns the next step, people have no time to practise or the task sits outside their normal workflow. My recommendation is to name an owner, choose one repeated task and agree how to assess quality, time and continued use. In my 2025 survey, 81% reported employer-provided generative AI, 25% reported using AI agents and 33% reported formal training. Agent use is different from total AI use. These findings describe respondents and do not explain why individual pilots fail. See the State of AI in HR data.
How long does AI adoption take?
The time needed depends on the tasks, systems, risks and support available. A 90-day plan can provide an initial review point: first let leaders practise and clarify rules, then test a few workflows, and finally decide which changes to keep. Review progress against observed use, quality and time spent. This is a planning example, not a promised timetable. For findings from Swedish HR respondents, see the State of AI in HR data.
Who owns AI adoption in an organisation?
Leaders should set priorities and assign an owner to each workflow, supported by HR, IT and other relevant functions. Teams need time to practise and a way to raise problems. In my 2025 survey, 3% framed AI primarily as an HR question, 39% as an IT question and 47% as both. That question captures perceptions; it does not establish who led each organisation’s AI work. See the State of AI in HR data.
What is the best way to implement AI change management in an organisation?
I recommend starting with leaders using the tools on their own tasks, alongside a few team trials. Agree which tools and data are approved, who reviews the output and when to evaluate each trial. Adjust training to people’s starting knowledge and the work they need to do. A review after 90 days can help decide what to continue, change or stop.
Make AI adoption happen
A keynote can introduce the choices facing your organisation. In a workshop, we practise on selected tasks and plan how to follow up. Tell me who will take part and what you want to work on.
Pilot vs. adoption: what’s the difference?
| A pilot | Continued use | |
|---|---|---|
| Goal | Prove the tool works | Change how work gets done |
| Scope | One team, time-boxed | Used in agreed workflows |
| After it ends | Decision to stop, adjust or continue | Built into daily workflows |
| Key signal | “It worked in the demo” | Used without being told |
| Next step | Assess the trial | Follow up continued use |
The five conditions at a glance
The data: see State of AI in HR: Sweden 2024 to 2025, with results on tools, training and AI policies.
Related guides
All guides on this site
Courses and learning
- Free AI course for HR, with a certificate of completion
- AI courses for HR: 14 free and paid options compared
- Free AI courses for HR: compare time, content and certificates
- AI upskilling for HR: build knowledge through real work
- An AI literacy playbook for HR: four phases of adoption
- AI dictionary for HR: 20 terms in plain English
- What is HR? Roles, responsibilities and everyday examples
Speakers and workshops
- AI keynote speakers for HR: 12 profiles compared
- How to Choose an AI Keynote Speaker (2026 Guide)
- Book an AI keynote speaker: process, price and what you get
- AI Keynotes
- AI Workshops
- AI Change Programs
AI in HR practice
- AI in Recruiting: A Practical Guide for Talent Teams
- 13 ChatGPT Prompts for HR That Get Results
- AI tools for HR: 20 options by task
- Which AI tool is best for HR? ChatGPT, Claude, Gemini and Copilot
- AI in performance management: preparation, not verdicts
- AI for people analytics and workforce planning
- Skills-based organisations: a practical starting guide
- Ethical AI in HR: a practical governance checklist
- MCP and AI connectors: a practical guide for HR
- AI agents for HR: lessons from my ChatGPT Operator test
- How to create a PowerPoint presentation with AI
- Will AI take HR jobs? Tasks, roles and what to do next
Leaders and adoption
- AI for HR: A Practical Guide to Adoption That Sticks
- AI for Leaders: A Practical Guide to Driving Adoption
- How to Run an AI Pilot That Doesn’t Die
- Getting started with AI: your organisation’s first 90 days
- State of AI in HR: Sweden 2024–2025
