Building Your AI Adoption Team
The AI council sets strategy. The adoption team makes it happen. Learn how to build the team, understand common barriers to AI adoption, and overcome each one.
What is an AI adoption team?
If the AI council is the board of directors, the adoption team is the project crew that builds the house.
The council decides “we’re deploying Copilot to 5,000 workers.” The adoption team figures out HOW: who gets it first, what training they need, how to measure success, and how to handle the people who don’t want to use it.
Without an adoption team, AI tools get deployed and then ignored. Licences get wasted. Employees get frustrated. The adoption team makes sure AI actually gets used — and used well.
Adoption team structure
Every AI adoption team needs these six roles. In smaller organisations, one person may wear multiple hats.
| Role | Responsibility | Why it’s critical |
|---|---|---|
| Executive sponsor | Removes blockers, secures budget, signals importance | Without visible leadership support, adoption stalls |
| Project manager | Manages timeline, milestones, dependencies, communication | Keeps the rollout on track and stakeholders informed |
| IT / technical lead | Handles deployment, integration, security, and support | Ensures the technology works reliably in the environment |
| Change management lead | Manages the people side: resistance, communication, culture | Technology changes fail without people changes |
| Training lead | Designs and delivers learning programmes | Users need skills, not just software |
| Champions coordinator | Recruits and supports peer advocates across the business | Champions drive adoption from the inside (covered in the next module) |
Exam tip: Adoption team vs AI council
The exam may ask you to distinguish between the AI council and the adoption team. The council is a governance body (strategy, oversight, approval). The adoption team is an execution body (deployment, training, support). They work together but have different mandates.
If a question asks “who approves an AI use case?” — the answer is the council. If it asks “who trains users?” — the answer is the adoption team.
Common barriers to AI adoption
Even with the best technology, adoption can fail. These are the six most common barriers — and how to overcome each one.
1. Fear and resistance
What it looks like: “AI is going to take my job.” Employees see AI as a threat, not a tool. They avoid using it or actively resist it.
How to overcome it:
- Communicate early and honestly: AI augments work, it doesn’t replace people
- Show concrete examples of AI making jobs BETTER (less admin, more interesting work)
- Involve employees in pilot programmes so they experience the benefits firsthand
- Address job security concerns directly at the leadership level
2. Skills gap
What it looks like: Users don’t know how to prompt effectively. They try AI once, get a poor result, and give up.
How to overcome it:
- Structured training programmes (not just a one-hour webinar)
- Role-specific prompt libraries (give people a head start)
- Ongoing learning: office hours, tips of the week, peer sharing
- Measure prompt quality alongside adoption rates
3. Data readiness
What it looks like: AI tools can’t find the right data, or find too much of the wrong data. Copilot surfaces outdated documents or content from the wrong department.
How to overcome it:
- Audit data governance BEFORE deploying AI (permissions, labelling, lifecycle)
- Clean up shared drives, SharePoint sites, and email archives
- Implement sensitivity labels and access controls
- Start AI deployment in areas with clean, well-governed data
4. Unclear ROI
What it looks like: Leadership approved AI but nobody defined success. Six months later, someone asks “was this worth it?” and nobody can answer.
How to overcome it:
- Define success metrics BEFORE deployment (time saved, quality improved, revenue impacted)
- Measure baseline performance first, then compare
- Report results monthly to maintain executive support
- Use both quantitative metrics (hours saved) and qualitative feedback (user satisfaction)
5. Shadow AI
What it looks like: Employees use free, consumer AI tools (ChatGPT, Gemini) instead of approved enterprise tools. Company data leaks to public AI services.
How to overcome it:
- Deploy approved enterprise AI tools quickly (don’t leave a vacuum)
- Create a clear acceptable use policy (what’s approved, what’s not)
- Make enterprise tools better than the free alternatives (integration, data access)
- Monitor for unsanctioned AI use and redirect users, don’t punish them
6. Leadership scepticism
What it looks like: Executives approved a pilot but don’t use AI themselves. Middle managers deprioritise AI training because “the boss doesn’t care.”
How to overcome it:
- Executive sponsor must visibly use and champion AI
- Share AI wins in leadership meetings and company communications
- Include AI adoption metrics in management KPIs
- Start with use cases that directly help leaders (meeting prep, email summaries)
| Feature | Root cause | Visible symptom | Primary tactic |
|---|---|---|---|
| Fear and resistance | Emotional — threat perception | Avoidance, complaints, low login rates | Communication + early involvement in pilots |
| Skills gap | Capability — don't know how | Poor results, 'AI doesn't work for me' | Structured training + prompt libraries |
| Data readiness | Technical — messy data environment | Irrelevant or wrong AI outputs | Data governance audit + cleanup BEFORE deployment |
| Unclear ROI | Strategic — no success definition | 'Was this worth it?' with no answer | Define metrics and measure baseline BEFORE launch |
| Shadow AI | Organisational — unmet needs | Consumer AI tools in the workplace | Deploy enterprise AI fast + clear acceptable use policy |
| Leadership scepticism | Cultural — lack of visible support | Middle management deprioritises AI | Executive sponsor models AI use visibly |
Scenario: Tomás builds PacificSteel’s adoption team
🔄 Tomás (Digital Transformation Lead, PacificSteel Manufacturing) needs to roll out Copilot to 5,000 workers across 12 factories and head office.
His adoption team:
- Executive sponsor: COO (factories report to her — she can mandate participation)
- Project manager: Tomás himself (he’s the DT lead)
- IT lead: Infrastructure manager (handles M365 deployment and security)
- Change management lead: Internal comms director (experienced with previous ERP rollout)
- Training lead: L&D manager (will build role-specific training tracks)
- Champions coordinator: An operations supervisor known and trusted across factories
Barriers Tomás anticipates:
- Fear and resistance — Factory floor workers worry AI will eliminate their roles. Tomás plans town halls at every factory, led by the COO, with a clear message: “AI handles paperwork so you can focus on production.”
- Skills gap — Many factory workers have limited tech experience. Training will be hands-on with real work scenarios, not generic AI tutorials.
- Data readiness — SharePoint has 8 years of ungoverned documents. Tomás budgets 3 months for data cleanup before the Copilot rollout.
His rollout phases:
- Phase 1 (Month 1-2): Head office (500 people) — tech-savvy, clean data
- Phase 2 (Month 3-4): Two pilot factories (800 people) — test factory-specific scenarios
- Phase 3 (Month 5-8): Remaining 10 factories — rolling deployment with lessons learned
Why phased rollouts beat big-bang deployments
Phased rollouts let you learn from each wave. Phase 1 reveals training gaps. Phase 2 reveals factory-specific challenges. By Phase 3, the adoption team has a refined playbook. A big-bang deployment to 5,000 people would overwhelm the support team and amplify every mistake.
Key flashcards
Knowledge check
Tomás discovers that factory workers are using free ChatGPT to write shift reports instead of the approved Copilot deployment. What barrier is this, and what's the best response?
Tomás is building PacificSteel's adoption team. He asks: 'Who on the adoption team is PRIMARILY responsible for addressing employee fears about AI replacing their jobs?'
Next up: AI Champions: Your Secret Weapon for Adoption — how peer advocates accelerate AI adoption faster than any training programme.