Why companies are afraid of AI (and why that's costing them)

Nobody told your team they were allowed to get it wrong.

That's why the tools aren't working. Why the pilots stall. Why the "AI-first" mandate quietly ghosts the all-hands deck. The problem isn't the model or the software. Nobody handed people permission to experiment, fail small, and try again.

And it's costing companies more than they think.

Ninety-five percent of enterprise AI pilots fail to move the P&L (MIT NANDA, 2025). The common read is "AI is overhyped." My take is that every company is handing tools to people who were never shown how to use them, then acting surprised when nothing changes.

"Make us AI-first"

A founder or an ops lead tells me they want to be "AI-first." Then, in the same breath, the real fear comes out. Our ambition is there, but we're behind. We're using zero agents right now. I know my business better than anyone, but I don't know the answer to these questions. They rarely say it outright, but underneath it, it's always the same thing. If my team can't figure this out, it falls back on me, and I don't have time to get pulled away from what I'm supposed to be focused on.

That fear is real. But it's not about the technology.

I've watched enough AI and automation agencies push productized rollouts and prepackaged "AI OS" systems. They rarely stick without rebuilding the business around them. Impressive on a sales call. Old and moldy three weeks later.

What works is less glamorous. It's the work most people don't want to do. Sitting inside a business for one week. Interviewing the team. Watching how work actually moves. I find it fascinating, honestly. I get to learn something new every time. I get to understand people, not just their workflows. That's where the real gold is. In the stuff you only find by listening. That's where the real friction shows up too. The gripes people are afraid to say out loud in a meeting.

Then I build around it and teach the people who'll actually use it, one on one. No stupid questions. I've never been shy about being the slow one in the room, so I ask them myself.

What this looks like in practice

One client runs their entire business through Notion. Somewhere along the way, that became the plan, one shared source of truth for the whole company. It's a smart call. But nobody ever circulated how to actually use it that way, so the plan and the practice drifted apart. That's why I'm not rolling out new software. I'm teaching people the tool that's been sitting in front of them the whole time.

Another client is a startup where the goal is faster decks, RFPs, and customer presentations. What surprised me wasn't an older team struggling to keep up. It was the opposite. The millennials and Gen Z on the team, the ones everyone assumes are already AI-native, weren't leveraging it in any way that moved the business. They were using it. They just weren't seeing how it could actually touch their day-to-day work.

The fear isn't the tools. It's permission.

Nobody tells people the first time they open ChatGPT or Claude. You don't need the perfect prompt. You don't need every integration set up. You need to think of it as a thought partner, something sitting next to you while you work through something you're not totally sure about.

Isn't that a completely different posture than "master this or get left behind"?

The employee is still the expert. The tool doesn't replace what they know. It removes the friction between knowing something and getting it done.

The other piece almost nobody gives people is permission to mess it up. Once someone understands they can walk back a bad AI-assisted attempt, that nothing's at risk, that it's not destructive, they start experimenting. And that's where real fluency comes from. Not a training manual. Experimenting.

Confidence is built, not bought

Half of all employees are already using AI tools their employer never approved (Software AG, 2024). The willingness was never the problem. Nobody needed convincing. They're already trying, alone, with no one to ask when they get stuck, because asking feels risky. Not risky like getting fired. Risky like looking like you don't have the answers, like you're not as sharp or as reliable as everyone assumes. So people quietly figure it out alone instead of raising their hand.

Companies that invest in structured training are 42% more likely to come out ahead on AI adoption (LinkedIn Learning, 2025). That's the whole gap between companies that get real value from AI and companies that just bought a lot of software. It's not the tools. It's whether someone taught the people holding them.

If any of this sounds familiar

If you're running a business or an agency, you already know it. Some of your processes are still manual. They could move faster. That's the conversation worth having.

I'm not going to pitch you a platform. I go into your business, find the friction, and teach your team to use what's already in front of them with real confidence. That's most of it, but not all of it. Once the foundation is solid, then we talk about what's next, not before. Bolt something shiny on top of a team that isn't grounded yet and it's just a band-aid. It looks good for a minute. Then it peels off, and you're right back where you started, just with another tool nobody's actually using. I like hard problems that deserve better solutions than "buy more software."

I'm Jon Sorrentino. This is the work I do. Reach out.