Our Methodology
Human-Centered AI Adoption
A human-centered approach to AI adoption — built to be customized for your team's unique culture, strengths, and mission. Like the scaffolding on a NYC building, the structure is the same but every application is unique.
Five Phases to Lasting AI Capability
The long-term risk of AI is not a lack of capability by the tools — it's a failure to have the human capability and institutional knowledge to use them effectively.
By prioritizing your team, you'll see better outcomes and greater engagement as your team embraces AI. This roadmap guides you through five phases — each designed to be customized like a scaffold, adapted to your unique culture and needs.
Phase 1
Explore
Map your current state, set a clear vision, and create basic guardrails.
Before your team can move forward with AI, you need to understand where you stand today. Phase 1 is about honest assessment — surveying your team, understanding organizational dynamics, establishing policy, and creating a safe foundation for experimentation. Your role as a leader is to be your team's AI sense-maker: translating the noise into a clear vision.
Key Activities
- 1Conduct a baseline assessment of current AI usage and attitudes
- 2Create a two-part policy: 'red lines' for prohibited use and a vision for AI success
- 3Establish foundational AI training and role-specific sandboxes
- 4Use a pull strategy — lead with invitation, not obligation
Frequently Asked Questions
Our team members are already using AI without permission. Should we punish them?
No. These are your early pilots, not policy violations. The booklet encourages treating discovered AI usage as early change champions rather than deviation from policy. Punishing them risks setting off an 'AI winter' in your organization as people fear retribution. Instead, redirect inappropriate usage from a place of appreciation for their experimentation and steer them toward safer practices.
Why do we need two policies instead of one?
A single policy tends to focus only on restrictions, which creates fear without direction. The roadmap recommends a two-part approach: the first section defines your 'red lines' — prohibited high-risk uses (your third rail). The second section sets a vision and goals for what successful AI adoption looks like, aligned to your broader strategy. Together, they create a cohesive narrative that explains what the organization will not do, what it intends to do, and why.
Why should AI adoption be optional through Phase 3?
Resistance is often a sign of uncertainty, not opposition. By keeping participation voluntary through Phase 3: Extend, you give team members a runway to observe wins and learn at their own pace. This pull strategy builds genuine engagement rather than reluctant compliance. When proven use cases become standard in Phase 4, most employees will already be ready to adopt.
Phase 2
Experiment
Run small, low-risk pilots. Document what's already in flight. Learn fast. Share wins and lessons learned.
Experimentation is where strategy turns into practice. This phase creates a safe space to explore, make small bets, and learn quickly. Leaders reinforce that experimentation within policy boundaries is encouraged, not feared. The goal is to move from theoretical understanding to practical confidence.
Key Activities
- 1Lead from the front — complete trainings and model active learning
- 2Appoint an AI Coordinator to track and connect pilot efforts
- 3Document current AI usage and establish directional metrics
- 4Expand leadership messaging and celebrate early wins
Frequently Asked Questions
What if our AI pilots fail?
The only failure is the one you don't learn from. The roadmap emphasizes celebrating early attempts even if not successful, and being consistent around messaging the importance of AI. Your greatest AI successes are likely from pilots your competitors didn't think of — so encourage experimenting now. Share lessons from pilots that didn't work to accelerate learning and lower the barrier for those who are hesitant.
Do leaders need to be AI experts?
No. Instead of striving to be a subject matter expert, the roadmap recommends modeling active learning for your team. Ask questions, show curiosity, and share both your insights and challenges. Learning is a skill many haven't practiced since school, so learning to learn as an adult is an important culture change you are well positioned to facilitate.
Why do we need an AI Coordinator?
As you discover pilots already happening — some planned, others informal — you need visibility. The AI Coordinator serves as the 'air traffic controller' for all AI efforts, tracking current and planned pilots, connecting related efforts, spreading best practices, and translating your vision into meaningful applications. This role helps prevent duplication and ensures lessons learned are shared.
Phase 3
Extend
Scale high-potential pilots, test metrics at scale, develop initial training and coaching plans.
Phase 3 transitions from early success with small pilots to scaling them for the broader team. Before scaling, review lessons from past change efforts and define clear go/no-go criteria. This phase is as much about learning how to scale as it is about increasing AI's impact.
Key Activities
- 1Assess and prioritize your portfolio of AI pilots for scaling
- 2Define clear success criteria and go/no-go decision points
- 3Keep adoption voluntary while communicating the path forward
- 4Establish an AI learning community with book clubs and promptathons
Frequently Asked Questions
How do we decide which pilots to scale?
Focus on use cases that frontline employees find most valuable and that clearly improve their jobs. Involve employees, your AI Coordinator, and leadership in selecting which pilots to scale. Leadership makes the final decision but should be transparent about why certain pilots were chosen — this transparency steers future pilots toward value.
What if a scaled pilot isn't working?
Don't be afraid to pull the plug. The roadmap warns against the 'sunk cost fallacy' — assess the project as it stands today without overly weighting everything invested. Document why it failed so you can revisit it if conditions change. Celebrate the attempt, capture lessons, and refocus energy on what's working.
How fast should we scale?
Avoid overwhelming your team by scaling too many pilots at once. Start with one or two, gather lessons, and adjust before expanding. Use short feedback loops — every two weeks — and consider the PDSA Cycle framework, which prioritizes tests of change over one big deployment. Scaling is as much about learning how to scale as it is about increasing impact.
Phase 4
Embed
Make proven practices part of the job. Update SOPs, onboarding, and performance expectations.
Embedding the first AI process into a team role marks a major milestone. The scaled pilot has proven its value and is ready to become part of how your team completes daily tasks. This phase turns successful experiments into standard practice — it should feel like evolution, not disruption.
Key Activities
- 1Formalize AI-enabled processes into SOPs, playbooks, and checklists
- 2Update training, onboarding, and refresher sessions
- 3Partner with HR to update job descriptions and hiring expectations
- 4Recognize success visibly and sustain through regular check-ins
Frequently Asked Questions
Why is partnering with HR critical at this stage?
If job descriptions don't reflect AI-related responsibilities, there's a disconnect between what people do and how they're evaluated. Adding clear AI language to roles, job postings, and onboarding sets expectations early, attracts AI-curious candidates, and prevents confusion over ratings and promotions. This conversation should have started in Phase 1 with the 'red line' policy — by now, HR should be well prepared.
How do we prevent AI usage from fading after embedding?
Embedding doesn't end when the process stabilizes — it transitions to monitoring and continuous improvement. Maintain visibility through regular check-ins between the AI Coordinator, leadership, and the individuals completing the process. Confirm the process is still being used as intended, capture refinements, and encourage enthusiastic users to serve as peer advisors for other teams.
Phase 5
re-Envision
Redesign roles and workflows to focus people on higher-value work as AI handles the routine.
This phase is about shaping the next version of your team. With hard-earned AI experience, you're ready to think beyond pilots and process improvements — redesigning work to deliver greater value. The roadmap does not recommend using this phase for reductions in force; your team's AI capability and organizational knowledge make them more important than ever.
Key Activities
- 1Reassess workflows and identify tasks AI can automate or simplify
- 2Co-create future-state roles with frontline employees
- 3Reimagine your value proposition for clients and stakeholders
- 4Cycle back through earlier phases as new AI capabilities emerge
Frequently Asked Questions
Should we use AI gains to reduce headcount?
The roadmap explicitly recommends against using this phase for reductions in force. Your team likely has more AI capability than someone from outside, along with invaluable organizational knowledge — they are more important than ever. Instead, use attrition and internal movement as opportunities to rebalance work, shifting roles toward strategic and customer-facing opportunities where humans excel over AI.
Is AI adoption ever 'done'?
No. As AI continues to become more capable, revisit earlier phases and repeat as needed — explore new capabilities, experiment with small pilots, extend what proves valuable, embed what works. Make reflection part of the culture: what did we learn, what surprised us, what will we change next time? This mindset keeps the organization adaptive instead of reactive.
Ready to Build Your Scaffold?
Take the AI Adoption Navigator to pinpoint your organization's current phase across the 5E framework — then get a personalized 90-day action plan you can export and share with your leadership team.