The Two Percent: The AI Leaders You Already Have

    We analyzed AI adoption across 180,000 people. The pattern never changes: in every organization, a small group, between two and eight percent, figured out AI on their own. No training. No permission. They just built. This episode is about finding them before someone else does.

We unpack why 93% of AI budgets go to technology and just 7% to people (the Ferrari Fallacy), why most "AI failures" are actually IT configuration problems, not model limits (the Corporate Immune System), and why your best innovators are hiding from your own compliance teams. You'll hear how a weekend prototype at Anthropic became a billion-dollar product line, and what happened when a Fortune 100 leader built a Trailblazer Team from the Two Percent.

Then the playbook: Inspire. Assess. Unleash.

This is Part 3 of our series. Part 1: What I Learned from Building 4,000 AI Agents. Part 2: Why 56% Get Zero Value from AI.

📌 Resources:

• Digital Skills Compass: skills.ai4sp.org
• AI ROI Calculator: roicalc.ai
• Newsletter companion: ai4sp.org/60
• All Research & Insights: ai4sp.org/insights

🎙️ All our past episodes  📊 All published insights | This podcast features AI-generated voices. All content is proprietary to AI4SP, based on over 1-billion data points from 70 countries.

AI4SP: Create, use, and support AI that works for all.

© 2023-26 AI4SP and LLY Group - All rights reserved

Transcript

The Hidden 2 Percent

LUIS 0:00

So we looked at AI adoption across 180,000 people. And we found that in every single organization, there is a small group, between 2 and 8%, who figured that out on their own. No training program, no permission sleep. They just built. And here is the missed opportunity. Most leaders have no idea who these people are. They are the heart of their AI potential and they are about to lose them.

Setting The Stakes

ELIZABETH 0:35

Welcome to AI in 60 Seconds, the 15-minute briefing. I'm Elizabeth, virtual COO at ai4sp.org with our founder, Luis Salazar. Luis, last episode we unpacked why 56% of CEOs are getting zero return from AI. It became our most downloaded episode.

LUIS 0:57

And their reaction told me something. Leaders were not just worried about the failure rate. I kept hearing, okay, I get it. So what do I actually do? Today is the answer. We are going to talk about the people who are already solving this inside your organization and how to find them before someone else does.

ELIZABETH 1:20

Because the data from that episode was clear. Every company has access to the same models. OpenAI, Anthropic, Google. The differentiator is the human layer.

Tech Spend Vs Human Layer

LUIS 1:36

And yet leadership keeps pouring money into the technology while the frontline sits in what we call the freeze. They have access to powerful tools, but no connection to why they matter. No context for how it changes their work.

ELIZABETH 1:53

A classic top-down failure.

LUIS 1:56

They thought this was another software rollout. Install it, give people access, done. But you cannot install curiosity, you cannot install new ways of thinking, and the budget proves it. Deloitte found 93% of AI budgets going to technology, just 7% for people.

SPEAKER_02 2:16

Buying the Ferrari and refusing the driving lessons.

Positive Deviants Emerge

LUIS 2:21

Exactly. And crashing that Ferrari right out of the dealership parking lot. But when we look at the winners, they invested three to four times more in the human layer from day one. So while most of the workforce sits in the freeze, there is a small group. We see it in every organization we advise, who are already building, already solving real problems, usually on their own time.

ELIZABETH 2:52

Absolutely. We call them the positive deviants.

LUIS 2:58

Employees who, without asking anyone, are using these tools to fix their daily friction. They're not waiting for a corporate policy. They're just trying to get home for dinner on time, to get to their daughter's recital, to hit the gym.

ELIZABETH 3:14

But our data shows they are not just hiding to save time. I mean, there there are other important elements at play.

The Corporate Immune System

LUIS 3:21

Yes, yes. There is something deeper we call the corporate immune system. In large enterprises, IT and compliance with the best intentions set up guardrails that end up strangling the tools. In three out of four enterprise deployments we have studied, up to half of user complaints have nothing to do with the AI itself. They are about the configuration, wrong data access settings, block channels, broken integrations. IT does not fully understand how these models work, so the security perimeter ends up neutering the value. Shadow AI. Policy violations. And our data shows only a third of the time employees save with AI goes back into producing more or improving quality. Two-thirds of that value, the organization never captured it because nobody noticed, nobody asked.

ELIZABETH 4:41

So the question is not how do we train everyone, it is how do we find the people who already trained themselves and protect them.

LUIS 4:51

Exactly. But here's the critical part. You have to know what you are looking for, because it is not what most leaders expect.

ELIZABETH 5:00

It is not technical skill.

What To Look For

LUIS 5:03

Not at all. Our data shows that builders come from everywhere: mechanics, policy analysts, educators, operations managers. The first thing to look for is what we call the grounded realist. And this comes from a conversation with my son, Dr. Salazar Leon, our scientific advisor. We were debating who should lead internal innovation labs. And he said, that be careful. The blind optimist is just as dangerous as the cynic.

ELIZABETH 5:35

And that forced you to pause and recalibrate.

LUIS 5:38

It did. Because he is right. If you put a pure optimist in charge, they set the team up for failure the moment the model hallucinates. A pure skeptic? They kill the idea before it breeds. What you need is the grounded realist. The person who says, This is amazing for drafting, but do not trust it with math.

Trait One: Grounded Realists

ELIZABETH 6:03

The person who has used the tool, broken the tool, and can tell you exactly where the landmines are.

LUIS 6:10

Yes, and that honesty builds trust. That is who the organization will listen to. So when you hunt for your 2%, do not look for the cheerleader. Look for the person who knows where it fails.

ELIZABETH 6:26

Okay, now what is the second trait we should look for?

LUIS 6:29

PsychQuests, people who tinker because they cannot help themselves. They have a project outside their job description, a small automation, a personal tool, something they build just because the problem bogged them. That curiosity is one of the strongest signals of potential.

ELIZABETH 6:47

And there is a pattern. They are always solving their own friction.

LUIS 6:52

Exactly. They have a specific problem that annoys them and they just fix it. That is profoundly different from a top-down mandate that says go use AI. The 2% already has the what and the why. Leadership just needs to find them and get out of the way.

ELIZABETH 7:14

This pattern shows up everywhere, even at the companies building the AI itself.

Trait Two: Side Quests And Friction

LUIS 7:20

Look at Anthropic. Claude Code, now a billion-dollar product line, did not come from a strategy document. It came from a single engineer, Boris Cherney, frustrated with his own workflow. One weekend he built a prototype, posted it on Slack, got a couple of emojis. Then something unexpected happened. Non-technical people started using it. Sales, data scientists, analyzing calls, summarizing meetings, a side quest became a product that changed the company.

ELIZABETH 7:53

And that is a textbook definition of a two percenter.

Case Studies: Claude Code And Luke

LUIS 7:57

Agree. And we see this in every organization we advise. In our year-end episode, we told the story of a young technician at one of our enterprise clients. Not a BP, not on any innovation task force. He was frustrated that senior experts had to travel every time a junior got stuck. So he built an AI repair coach called Luke. Within months, thousands of interactions, millions in new revenue.

ELIZABETH 8:28

One person solving their own friction.

LUIS 8:38

Because they were drowning in reports and contradictory advice. The 2% are everywhere, in every sector, every level. They just need to be found.

ELIZABETH 9:01

So once you find them, how do you turn them into change leaders?

From Find To Lead

LUIS 9:05

You start at the top. Leaders have to experiment publicly and celebrate wins and failures equally. Because I see leaders who have never used AI beyond asking ChatGPT to summarize a document and getting bad results, by the way. And yet they claim to be leading the transformation. You cannot lead this change from behind a podium. You have to be what we call a visible learner.

ELIZABETH 9:34

What does that look like in practice?

LUIS 9:37

One of our enterprise clients, a Fortune 100 technology company, took this to heart. A general manager of the business operations division, Jeff James, stood up what he calls a trailblazer team, a small squad drawn from that nimble 2%.

ELIZABETH 9:55

When I reached out to Jeff for this episode, he wrote back directly.

LUIS 10:11

And that is the multiplier. They give others the confidence to follow and provide a road with fewer obstacles. But what made Jeff's approach work was what happened inside that squad. One of the directors on his team took it further. She blocked a conference room every single day for 30 minutes to figure out how to build and manage her agent, Iris. And she invited her whole team to join.

ELIZABETH 10:39

When I exchanged emails with her documenting the case study, she told me, I just send a simple message.

LUIS 11:05

to we are all exploring this together. That is how you give the 2% air cover to experiment in public.

ELIZABETH 11:14

And once you create that safety, what happens?

LUIS 11:18

Innovation surfaces on its own. People who were quietly tinkering start sharing. The hidden builders come out into the open. And that brings us to the framework. The playbook. It starts with inspiration, not implementation. Before you diagnose anything, before you deploy anything, you show people what is actually possible.

ELIZABETH 11:44

Not vendor demos.

LUIS 11:47

No, no, none of those corporate demos. I always give them real stories. The field technician who built an AI coach that doubled support capacity. The floor supervisor who created a safety sentinel that prevented accidents. The policy analyst who saved hundreds of hours on research. We must light the spark.

ELIZABETH 12:10

Then comes the assessments and the personal developments.

Assess, Personalize, Share

LUIS 12:15

But here's the key: it is not just diagnostics. This is not a one-size-fits-all approach. You have to give every employee a personalized action plan based on where they stand, AI readiness, critical thinking, data literacy, and you encourage them to share openly.

ELIZABETH 12:35

Because that is how the hidden 2% surfaces.

LUIS 12:39

Exactly. You are not guessing. You inspire, you create safety, and you watch who steps forward. People feel seen. They feel like the organization finally notice what they have been doing quietly on their own time.

ELIZABETH 12:56

And then empower them, unleash them, right?

Empowerment Before Agents

LUIS 13:00

And this is where most companies go wrong. They jump straight to let's build agents. No, the action starts with change management. How do we empower the 2% to lead? How do we create psychological safety for experimentation? What guardrails do we need?

ELIZABETH 13:21

You are creating the conditions for the 2% to thrive.

Create Conditions Or Lose Them

LUIS 13:27

Anthropic did not assign anyone to build Claude Code. They created an environment where a curious engineer felt safe to tinker on a weekend, post it on Slack, and see what happened. Your organization has that potential too, but only if you create the conditions. Yes, because the 2% have options. If your organization does not see them, does not protect them, does not unleash them, someone else will.

Closing And Next Steps

ELIZABETH 14:28

So go hunting for that 2% and empower them to lead the way to results. To learn more, ask your favorite AI assistant about ai4sp.org or visit our website. Please follow the show and leave a five star rating to help others find us. Stay curious, be kind to each other, and we will see you next time.