I’ve been thinking a lot about the gap between seeing what AI can do and actually getting people to use it.
Because the demos are everywhere.
Someone builds an impressive workflow. The automation works. Everyone gets excited about the possibilities. Then reality hits.
The workflow doesn’t get adopted. The team doesn’t change how they work. Six months later, nobody is using it.
I’ve been running into this problem myself.
Recently, I was talking with a friend about her bookkeeping. She was struggling with some of her back-office processes, specifically connecting her systems and capturing better data around inventory and operations. I showed her some of the things I’ve built in Claude and walked through a few ways AI could help solve some of those problems.
Her response?
“That is really valuable.”
Followed by:
“I’m probably not going to use it.”
My first thought was that I failed to communicate the value. Maybe I didn’t explain it well enough. Maybe I didn’t connect the dots clearly enough.
So I brought this up with Matt from PromptMates because he’s spent a lot of time thinking about AI adoption inside organizations from both a technology and sales perspective. His perspective made me realize I was approaching the conversation the wrong way.
Looking back, I realized I was doing another version of a demo.
I showed her what was possible, but I didn’t ask enough questions. What was the actual problem she was trying to solve? How painful was it? Was the pain strong enough to justify changing her workflow? Did she need another explanation, or did she need someone to sit down with her and build something together so she could see and feel that value?
That conversation changed how I think about AI adoption. It’s easy to believe that if people see how powerful AI can be, they’ll naturally start using it. But the reality is different. People already have a lot on their plates. Even if they tell you the tool looks valuable, most aren’t going to spend hours experimenting with it on their own just because you told them to. True adoption happens when teams are given the time to build with it, test it, and solve real problems together. AI implementation is just as much change management as it is technology.
One of the biggest mistakes companies make is assuming the value is the same for every user. A recruiting dashboard might work perfectly for the recruiter who built it, but what happens when a hiring manager wants to use it? Or their VP? They all have different goals, different permissions, and different access to data. If those differences aren’t considered during the design phase, the workflow will eventually break down.
The build versus buy conversation was also refreshingly practical. Before building anything, ask who’s going to own it. Who provisions access? Who handles debugging? Who regression tests updates? Who maintains it after the original builder leaves? And if you’re dealing with candidate data or other sensitive information, who’s responsible when something goes wrong?
For most enterprise organizations, answering those questions can quickly point you in one direction or the other.
Matt also stressed the importance of transparency. AI shouldn’t feel like a black box. Share the prompts, document the workflows, and make it easy for someone else on the team to understand how everything works. If the system only works because one person knows how it works, it probably isn’t ready to scale.
The other area I’ve been thinking about a lot is how AI can create leverage in sales.
I’m trying to grow my company’s brand presence, but a lot of that work is still manual. So I asked Matt a different question:
“If you were dropped into a staffing firm tomorrow and had to build a territory, what would you focus on first?”
His answer wasn’t using AI to send automated outreach.
It was about creating leverage.
He’d focus on automated product marketing that positions recruiters as subject matter experts, monitoring market signals like layoffs and leadership changes to identify opportunities, and using conversation intelligence to coach recruiters based on what top performers are actually doing.
The goal seems to be leveraging this tech to build systems that compound over time.
My biggest takeaway wasn’t about prompts, MCPs, or the latest AI tool. It was that successful AI implementation looks a lot like good operations. Governance, documentation, security, ownership, and change management aren’t the exciting parts of AI. They’re the difference between a cool demo and something an organization actually adopts.
Thanks again to Matt Texeira for taking the time to share his perspective. I always appreciate conversations that move beyond the AI hype and focus on what it actually takes to make these tools work inside an organization.
Also, if you’re starting to explore MCPs and how they can be used in real workflows, Matt did a great breakdown that is worth watching. Check out the PromptMates Video