When AI Is the Amplifier: Rethinking How We Lead Change and Adoption

 

This Isn’t a Platform Swap

If an organization moved from Microsoft Office to Google Workspace, legacy approaches to change and adoption would earn their keep. The destination is known. The mapping is apples-to-apples, documents to documents, spreadsheets to spreadsheets, mail to mail. Communications, training, and a champions network usually get people across.

AI, agentic AI, and AI-native work are nothing like that.

We’re not replacing one tool with another. We are literally reimagining the ways in which organizations get things done. What roles do we need? What do workflows look like when agents are participants rather than features? How do people and agents collaborate? How do agents collaborate with each other? In an AI-native posture, we can even use AI to find the problems worth solving — and to help build the solutions. The destination isn’t fixed; it’s an evolution shaped by capabilities that are changing weekly.

When the destination is emergent, a roadmap is the wrong tool. What we need is a better compass. We need something different, not as a replacement for discipline, but as a different kind of discipline.

 

A Quick Map: Cynefin

Cynefin is a sense-making framework that sorts situations into domains and tells us what kind of response actually fits each one. The short version:

• Clear: cause and effect are obvious. Sense → categorize → respond. Best practices work.
• Complicated: cause and effect exist but require expertise. Sense → analyze → respond. Good practices and experts work.
• Complex: cause and effect are only knowable in retrospect. Probe → sense → respond. We learn our way forward through safe-to-fail experiments.
• Chaotic: no discernible cause and effect. Act → sense → respond. Stabilize first, make sense later.

The Office-to-Google migration lived in Clear, with some Complicated edges. Legacy approaches to change and adoption serve that domain well.

Most of what AI adoption asks of us today lives in Complex, with sudden visits to Chaotic when a new capability lands and resets assumptions. That means our default move can’t be “plan the rollout and manage the change.” It has to be “probe, sense, respond” — run small, safe-to-fail experiments, watch what emerges, and amplify what works.

If our change approach doesn’t match the domain we’re actually in, we’ll do the right things in the wrong way and wonder why adoption stalls.

 

Human-Centered AI: Working, Thinking, Building, Adapting Together

Cynefin tells us how to move. Human-Centered AI (HCAI) tells us what we’re moving toward: a way of working in which humans and agents collaborate as first-class participants, with new UX and AX (agent experience) thinking woven through.

Practically, that’s four “togethers” worth designing for:

• Working together: humans and agents share workflows, with clear handoffs, observability, and the ability for either to flag uncertainty.
• Thinking together: agents extend our reasoning rather than replace it; people retain meaningful judgment, especially where stakes, ethics, or ambiguity are high.
• Building together: teams use agentic capabilities to identify problems and prototype solutions, shortening the loop from idea to evidence.
• Adapting together: the system (people and agents) learns from each cycle. New capabilities get absorbed continuously rather than batched into transformation programs.

This is where UX and AX matter. We’re no longer just designing how a person uses a tool. We’re designing how a person and an agent, or several agents, share work, context, and accountability. That’s a new design discipline, and it belongs at the center of adoption, not on the side.

 

What This Looks Like in Practice

A team we’ve recently been working with wanted to lift learning across the organization. They already had an embarrassment of content, Skillsoft subscriptions, thousands of courses, and the open internet. And, yet, adoption stayed low. People weren’t short of options; they were lost in the abundance.

We started with people, not technology. Interviews and focus groups surfaced hopes, fears, and goals — and made it plain that different people and different roles needed different things. From there, we ran tailored in-person sessions shaped around their mindset, their goals, and their actual work. Only then did we build the AI-native layer: a deliberate learning architecture, curated content, and a small set of agents. One agent helps each person access the curated material, mapped to their role, goals, and work. A few more support a curatorial team that keeps the library fresh and aligned. Thoughtful architecture, thoughtful agents.

Notice what changes here:

• Learning shifts from creation to curation. The half-life of content stops mattering as much.
• Each learner gets a path shaped to their role and work, not a cohort average.
• The agents themselves become a live example of human-agent collaboration; people see HCAI by using it.
• The team running it learns more from a quarter of operating the agents than they would from a year of designing a course.

That’s a complex-domain move: discovery first, then deliberate experimentation — instrumented, safe to fail, and designed to teach the organization something it couldn’t have predicted from a plan.

A few principles tend to travel with this kind of work:

• Psychological safety and radical candor. People will only probe honestly if it’s safe to surface what isn’t working, and direct enough to act on it.
• Permission to play, with guardrails. Sandboxes, clear data boundaries, and lightweight governance that enables rather than gates.
• Safe-to-fail, not fail-safe. Design experiments so a bad outcome teaches us something cheap, not so nothing bad can ever happen.
• AI champions, distributed. Not a central team gating access, practitioners across functions who pull capability into their own work.
• Anti-fragility over robustness. We want an organization that gets better under stress and surprise, not one that merely survives it.
• Personalized learning by curation. Pull from the world’s best material; shape it to context.
• Continuous adaptation, not transformation. There’s no end state to land. There’s a maturing capability to tend.

All of this comes back to the amplifier. AI turns up the volume on whatever the organization already is and is doing: strategy, skills, ways of working, the capacity to learn and innovate, and the culture all of that produces. Culture isn’t an input. It’s an output of the system, seen most clearly in the least desirable behavior we’re willing to tolerate. AI will amplify it either way.

 

So, What Do We Call It?

We’ve been careful not to lean on the legacy vocabulary here. Honestly, we’re not sure “management” is the right word at all. Management implies control over a known thing. What we’re describing is closer to leadership in conditions of genuine uncertainty — discovery, sense-making, and disciplined experimentation alongside the people doing the work.

Some names worth chewing on:

• Adaptive Change Leadership
• Continuous Adoption & Thriving
• Adoption, Discovery & Adaptation
• AI-Era Change Leadership
• Human + Agent Adoption

We don’t think the label is the point. Whatever we call it — x, y, or z — what matters is that we recognize this work as strategic, fund it accordingly, and stop treating it as a comms-and-training appendage to a technology rollout.

 

What We’re Inviting

We’re not arguing that legacy approaches were wrong. We’re arguing they were built for a domain we’re no longer mostly in. The invitation is simple, and it sits with us as leaders:

• See the domain we’re actually in. Notice when work is Complex or Chaotic, and stop applying Clear-domain playbooks to it.
• Probe deliberately. Fund small, instrumented, safe-to-fail experiments. Amplify what works; learn loudly from what doesn’t.
• Design for humans and agents together. Treat HCAI and AX as core, not garnish.
• Invest in the amplifier’s substrate. Strategy, skills, ways of working, psychological safety. AI will turn the volume up on all of them.
• Tend the capability, continuously. This isn’t a transformation with a finish line. It’s a practice.

We’re all watching capabilities arrive faster than our org charts can metabolize them. Some of us are sprinting; some are practicing strategic patience with focused bets. Both can be right; what matters is that the way we’re leading change matches the kind of change we’re actually in.

We’d love to hear what you’re calling this work, and what’s working.

 

 

Photo Credit: Pawel Czerwinski | Unsplash