I’ve watched this happen more times than I can count. A company buys AI licenses for the whole team. Everyone gets a login. Six weeks later, maybe three people are actually using it.
So when I came across a post from Jeremy Bennett breaking down a conversation with Ruben Hassid, it clicked right away. The original poster runs a series featuring LinkedIn’s top creators, and this episode covered Ruben’s newest launch, GPC. Its whole focus is helping companies actually adopt AI. What grabbed me wasn’t the launch itself, though. It was the way Ruben thinks about why adoption fails in the first place.
Here are the big questions this episode answers.
What’s the real problem with AI at work?
Most people assume it’s access. It isn’t.
Ruben’s take is that companies already have AI. They already know the tools exist. But having the tools and using them are two very different things.
He frames the bigger challenge as a human one. It’s a change management problem, not a tech problem. Meaning, the hard part is changing how people work, not picking the right software.
That reframe alone is worth sitting with. If your rollout plan is mostly about picking vendors, you’re solving the easy half.
So what does success actually look like?
According to the expert, you need to get an entire company to do three things:
- Try AI.
- Get comfortable using it.
- Keep using it without being pushed.
That last one is the tricky part. Anyone can force a team through a training session. Getting people to come back on their own, on a random Tuesday, because they want to? That’s the real finish line.
Why do some people jump in while others stall?
This is my favorite part of the post. Ruben splits people into two groups: “clickers” and “non-clickers.”
Clickers naturally explore new technology. Hand them a new tool and they’ll start poking at it right away. You don’t need to convince them.
Non-clickers need more support. And that’s totally fine. According to Ruben, they need three things:
- Permission to try.
- Small wins to build confidence.
- Enough practice to explore alone.
I love how simple this is. Most AI training assumes everyone is a clicker. It throws a tool at people and expects curiosity to do the rest. For a big chunk of any team, that just doesn’t work.
How do you move a non-clicker forward?
Ruben compares it to riding a bike.
You support someone in the beginning. You hold the seat, you jog alongside them. But eventually, they need to ride alone.
The goal of AI support isn’t to make people dependent on help. It’s to get them confident enough to let go of the seat.
If you’re leading a team, here’s how I’d put this into practice based on Ruben’s framework:
- Say it out loud: people have permission to experiment with AI, and mistakes are expected.
- Pick one boring, repetitive task per person and show them how AI handles it. That’s the small win.
- Pair a clicker with a non-clicker for a couple of weeks.
- Step back gradually so people start exploring on their own.
What does a real adoption plan look like?
Per the post’s author, that’s the approach behind GPC. The process starts by assessing where a company currently stands with AI. Then it builds an adoption roadmap around three questions:
- Which people could benefit from AI?
- Where could it improve their daily work?
- Who needs help becoming confident with it?
You can run these questions yourself, even without outside help. Grab a list of your team, jot down one daily task per person that AI could speed up, and mark who’s a clicker and who isn’t. You’ll have a rough roadmap in under an hour.
Isn’t this really about cutting jobs?
Ruben was clear on this one. The goal isn’t to cut people’s jobs. It’s to help people do more with AI.
I think this point matters more than anything else in the post. Non-clickers often hold back because they’re quietly worried. If AI feels like a threat, nobody’s going to practice with it. If it feels like a boost, they’ll actually lean in.
My takeaway
The clickers vs. non-clickers idea is going to stick with me. It explains why so many AI rollouts stall, and it gives you a clear path to fix them. Support first, independence later, and never lead with fear.
The original post also asks a fun question: is AI actually easier to use than people think? I’d love to see how folks answer that. Check out the full LinkedIn post to read the whole breakdown and join the conversation in the comments.