AI-Native SAFe

June 26, 2026

The Problem with AI Isn't the AI

# AI

# SAFe Summit

Most teams are paralyzed by infinite possibility. Here's how to break through.

This keynote from SAFe Summit Denver 2025 is available as part of your SAFe Community membership. Below, we break down Alison McCauley's core framework and give you a way to use it with your team this week. The full talk is embedded at the end of this article.

The Problem with AI Isn't the AI

Sometimes the hardest part of having infinite capability is knowing where to start.

Your team is a microcosm of the larger pattern. A few people are deep into AI—building agents, testing workflows, experimenting daily. Most are casual users—summarizing meetings, drafting emails, occasionally asking ChatGPT for help. And a handful haven't really engaged at all.

Everyone knows it matters. No one knows what to do about it.

This isn't a training problem. It's not a tools problem. And it's definitely not a "people just need to get on board" problem.

It's an alignment problem.

In her keynote at SAFe Summit Denver, Alison McCauley—who's spent 30 years at the intersection of emerging technology and human behavior—named something most organizations are experiencing but haven't quite articulated:

AI gives us a vast landscape of possibility. And that landscape is paralyzing.

She calls it the paradox of choice. When you sit in front of AI, you often don't know how to engage. You freeze.

The teams that break through don't solve this with better prompts or fancier models. They solve it by going analog first.

The pattern that's repeating

McCauley opened with a reminder: every transformative technology arrives with both fear and excitement.

When electricity was introduced, it triggered the "electric wire panic." President Benjamin Harrison refused to touch the light switches in the White House. His staff did it for him.

ChatGPT reached a million users in five days. A hundred million in two months. The fastest consumer tech adoption in history.

Then came the whiplash.

Governments blocked it. Schools banned it. The Italian government took it away from citizens for a month.

Less than three years later, the U.S. government struck deals to give it to the entire federal workforce for $1. There's now a K-12 initiative to have kids use AI to solve complex global problems.

That's the pattern: overestimate in the short run, underestimate in the long run.

But here's what's different this time.

Why AI breaks the playbook

Everything we've built in organizations has been built on traditional software. Precise. Predictable. Debuggable.

AI never delivers the same answer twice.

That's not a bug. It's the nature of generative systems. But it means every workflow, every role definition, every quality control process we've designed assumes predictability.

AI doesn't offer that.

McCauley put it plainly: "We have built our systems on the assumption of cognitive scarcity. AI moves us into cognitive abundance. Some of those systems are going to break."

She's not talking about distant future disruption. She's talking about right now.

Voice authentication? Defeated. CAPTCHAs? Agents click through them casually. The systems we built to verify humanness are dissolving.

Meanwhile, a startup McCauley spoke with has 30 "team members." Six are human. The rest are AI agents—with roles, responsibilities, decision rights, and communication channels.

That's not science fiction. It's Tuesday.

The competitor you haven't met yet may not have a big team. They might just have better orchestration between human and machine.

So the question isn't whether AI will change work. It's whether your team will be ready when it does.

The unlock: alignment before adoption

Here's where McCauley's background in early change management becomes crucial.

In the 1990s, she worked on one of the first CRM projects—pre-Salesforce, when getting sales teams to put notes in a spreadsheet instead of their pockets felt like heresy.

She traveled the world for that project. And the world hated her.

It was only when her team discovered three principles that adoption finally moved:

Meet people where they are

Demonstrate value that deeply matters to them

Patiently guide them toward that value

Those principles still work. But most teams skip straight to step three.

They roll out tools. They run training. They hope people will figure it out.

The teams that break through do something different.

They start with a problem worth solving.

The analog-first workshop

McCauley walks leadership teams through a deceptively simple exercise.

No laptops. No AI tools. Just whiteboards, markers, and an unreasonable number of Post-its.

The questions are introspective:

What's a recurring problem we've never cracked?

What bottlenecks slow us down?

How could we leverage our core competency to do something totally new?

If only we could _____, then we could _____.

This is imagination space. Dream space.

You might have asked these questions before. But everything changed in the last few years. The constraints shifted. What was impossible is now merely hard.

Once teams generate a library of possibilities, they filter through significance, motivation, strategic alignment, and frequency. They narrow to one or two focal points.

Then—and only then—they bring AI into the conversation.

They use it to co-create a point of view on how AI might help. Not in the abstract. For this specific problem. With this specific team.

What comes out of this process isn't just a use case. It's:

A shared focal point that cuts through the paradox of choice

Cross-functional alignment between business, data, and tech teams

Motivation to persevere because the problem is significant enough to warrant the discomfort of learning

McCauley calls this process an accelerant. You can take a first pass in a few hours.

Then you rinse and repeat.

What alignment enables

Once you have that focal point, the path forward clarifies.

McCauley demonstrated what she calls "deep thinking with AI"—using it as a strategic thought partner, not just a productivity booster.

She recorded herself playing the role of a McKinsey partner at risk of AI disruption. In 30 minutes, she:

Ran deep research across 200+ websites using Gemini

Generated two comprehensive reports (28 and 30 pages)

Synthesized insights across both using ChatGPT

Created discussion materials and slide decks using Gamma

Drafted leadership communications using Claude

In the past, that would have taken days—or been delegated to someone junior.

The key wasn't the tools. It was the expertise to know what questions mattered and what answers were worth keeping.

AI amplifies expertise. It doesn't replace it.

Your experts are more valuable than ever. They're the ones who can tell the difference between convincing output and accurate output. When they're boosted by AI, the result is a 3x productivity gain across a wide range of activities.

And when teams are aligned around problems that matter, those experts have a clear target for their amplified capability.

The identity question beneath the strategy

This is where McCauley's talk shifts from tactical to existential.

She asks: What do we do with the time AI gives back?

Before the Industrial Revolution, doing laundry took 8-12 hours a week. Now it's an hour. Tomorrow, maybe not at all. (She showed a humanoid robot from Figure folding clothes.)

In 1966, the BBC interviewed school children about the year 2000. They predicted automation would make life boring—that everything would be the same, and there wouldn't be enough jobs.

That's not what happened.

Could it be that when we're surrounded by synthetic everything, we choose to invest our freed capacity in what makes us uniquely human?

McCauley suggests that in a world where everyone has access to the same powerful AI, the real differentiator becomes human connection.

Not as a soft skill. As a strategic asset.

Better patient outcomes. Deeper customer relationships. Complex ecosystems that require trust to navigate. Immersive forms of entertainment that demand genuine human creativity.

The business case for human connection gets stronger, not weaker, as AI capability becomes abundant.

But that requires a choice.

Organizations will face a question: Do we capture productivity gains for short-term shareholder returns? Or do we reinvest that capacity into building something that wasn't possible before?

Teams will face a parallel question: Do we delegate our thinking to AI and let our capabilities atrophy? Or do we use AI to push ourselves to think harder and work better?

These are identity questions, not process questions.

The learning infrastructure you need

McCauley offered three practical accelerants for teams trying to build momentum:

1. Activate the passionistas

Find the people who are already experimenting—even in their personal lives. They hold valuable knowledge. You can't keep up with AI innovation alone. No one can. Your success depends on building a learning network.

2. Create forums for exchange

Run no-code hackathons where people "hack with their words." Host show-and-tells. Buy lunch and let people share what they've learned. Make it social and low-stakes.

3. Tell stories about the journey

Don't just showcase the wins. Talk about the bumps along the way. What didn't work? What are you still struggling with? That's where the real learning lives.

The meta-lesson: AI fluency isn't something you train into people. It's something you cultivate through shared experimentation around problems that matter.

What Monday morning looks like

Here's the exercise you can run with your team this week.

Time required: 30-45 minutes Materials: Whiteboard or shared doc, nothing fancy Participants: Your immediate team (5-10 people max)

The prompt:

"We're not talking about AI yet. We're talking about our work. I want to hear about:

One thing that's been frustrating you for months

One process that feels slower than it should be

One thing you wish we could do but haven't had capacity for"

Give people 5 minutes to think. Then go around the room.

Don't solve anything yet. Just listen and capture.

Once you have 8-12 items, ask the group:

"If we could only focus on one of these—the one where progress would unlock the most value or relief—which would it be?"

Vote. Pick one.

Then ask:

"What would 'solved' look like? What would we be able to do that we can't do now?"

Write that down.

That's your focal point.

Next step: Take that focal point into a conversation with AI. Not to get answers. To get questions.

Ask it: "If you were helping a team solve [problem], what would you want to know first? What factors should we consider? What have other teams tried?"

You're not looking for a solution. You're looking for a better frame.

Do this once. See what shifts.

Then decide if it's worth doing again.

Watch the full keynote

McCauley's complete talk—including live demos of AI agents, her analog workshop process, and the full "cognitive abundance" thesis—is embedded below. It's 47 minutes that will reframe how you think about AI adoption.

The talk moves from historical context to hands-on demonstration to strategic implication. If you're a team lead trying to figure out where to start, it's worth the investment.