How to get your team to ACTUALLY dive into AI

Most people know AI can make them more productive. Most people still aren't using it well.

Pete Larkin has heard every reason why. As co-founder of UXM and someone who has built AI deeply into both his work and personal life, he sat down with Paxton Gray on The Campaign to go through the most common barriers to AI adoption, one by one, and talk through how to beat them.

Before the roadblocks, though, Pete had something to say about why the effort is worth it.

"Instead of just running at 150, 160% capacity with these new superpowers, I'm able to reinvest into the things that I enjoy the most," he said. At the organizational level, he sees something bigger than efficiency gains: "It's not just how do we take AI and kind of fit this into a box of how we already do things, but how can we fundamentally step back and rethink the way that we do this work and reinvent it."

That mindset shapes everything that follows.

Barrier #1: Time

By far the most common answer in Paxton's informal survey of 40 people: I don't have time to learn this.

Pete's take is that time is a framing problem.

"Think about planning for retirement," he said. "It can be hard to do that upfront, but then later you get to enjoy it. And it's very similar with how we invest into AI. You don't see the gains immediately. You see it over time. And as you build out your new automations and workflows, those gains come back in multiples."

The first step is deciding you believe the investment is worth it. The second is practical: put it on your calendar.

"I time block. I put it into my calendar. I'm on average around 45 minutes during the work day," Pete said. He also acknowledges he's a little unusual: "It's fun for me. So I invest personal time in this. When my spouse is tired and going to bed, AI school is starting." For everyone else, he recommends finding a quieter part of the day and making the block recurring. "If you don't actually put it on your calendar and time block it as a recurring thing, it's going to be really hard to build out your AI fluency."

Where to start once the time is blocked: Pete's first recommendation is the free courses from Anthropic and OpenAI. "Go do the free courses about Claude. They're great content. Engaging, very professional, well-organized." On YouTube, he follows Nate Hurk, who runs an agency called True Horizon. But there's one rule for any content you consume: "You can't just be watching it. You have to open your computer and follow along. You have to be doing it as you're learning it, or it won't stick."

For leaders: The chicken-and-egg problem is real. Employees feel pressure to perform with skills they haven't built yet. Pete's advice is blunt: "You can't expect your employees to wake up one day and just magically have AI skills. You have to invest in the tools. Give them budget. But then you also have to invest in enabling them to have time to learn these skills during work time. You can't expect it to be during their personal time. It's just going to burn them out."

He suggests building it into the structure of the week. "Have your AI hour every day. You could even say, hey, full company from two to three, this is AI hour." Beyond that, bring it into one-on-ones and make it an expectation, not just a suggestion.

Barrier #2: Cost

The fear of racking up huge bills comes up more than it should, Pete said.

His answer is simple: start with Claude Pro.

"To me, it's hands down, no question about it: Claude. That's where you should put your money. Go to the pro account, pay the $20 a month. If your company isn't paying for it, go to the $20 a month account. It's well worth the investment."

From there, build out your capabilities and layer in additional tools as the need becomes clear.

On the credit usage fear: "You don't have to worry too much about credit usage unless you're building via API or getting into Claude Code and running through tons and tons of tokens. And if you're running through that many tokens, it's probably because the work you're doing merits it." The concern is largely overblown for anyone focused on building their skills.

Barrier #3: Hallucinations and Trust

Pete's answer here is that trust is earned by learning how the tool works, not by assuming it either does or doesn't get things right.

"The models continue to get better and better. The frequency of significant hallucinations is dropping," he said. But the bigger point is knowing where AI struggles. "Math, for example. When I'm doing sometimes pretty simple calculations, AI just struggles. So I know in what situations I need to be extra diligent in reviewing the results."

That diligence should apply across the board. "Before you even use AI, you need to think about: is this the right use case? Is this the right time and place?" If you're producing customer-facing marketing materials, review them. "Don't just trust it on its own to be high quality. Expect to have some edits and adjustments."

The rule is consistent: good input produces better output. Bad input produces what Pete calls AI slop.

Barrier #4: Dependency

This one came up less often in the survey, but Pete finds it the most interesting to think through.

His analogy: churning butter.

"How many people do you know that know how to churn butter? It was a very important skill at some point. Advancements in technology make certain skills irrelevant or unnecessary." When it comes to memorizing Excel formulas, for instance, he's comfortable letting those go. "If the tools can do them for me, I think it's okay if it goes the way of the butter."

But critical thinking is different. That's what people are really worried about when they raise the dependency concern, and Pete agrees it deserves attention.

"I think when you're using AI as a thought companion and to help you enhance your current skills, it's not a dependency where you stop thinking. It can actually help you grow and evolve the way that you think." The caveat: you have to engage with the process, not just collect the output. "Don't just tell the AI to give you the answer, but to walk you through it and help you understand how it got there. So it becomes a learning tool."

Barrier #5: Ethics

The ethical questions around AI are real, and Pete doesn't dismiss them. But his take is grounded in a clear principle: don't try to pass AI off as human when it matters.

"I think we should be transparent and open about how we present AI work," he said. He draws a sharp line at deception. "Imagine you call a customer support line and you're trying to get answers and you get on the line and it's AI talking to you. And if you ask, 'is this AI?' and it lies to you, I think that's super unethical."

He also has a strong view on cold outbound. "To have an AI call a human to try to sell something, all that says to me is that you don't care enough about me as a customer to send a real human to talk to me. I think it's bad taste."

At the same time, Pete sees AI as a legitimate creative medium, similar to how photography was once dismissed by painters. "It's a new medium. But in my opinion, it's very much a form of artwork."

The TIME Framework

Paxton Gray added a frame that ties the barriers together: the TIME acronym, built around the insight that time is the number one thing blocking adoption.

T is for Taste. Strong taste means knowing what drives action and how to persuade. Without it, you get volume without impact. "If you have no taste, that's AI slop."

I is for Impact. The goal is not more deliverables. It's more impact on the bottom line.

M is for Matching. Develop the instinct to know which tasks belong with AI and which belong with human judgment.

E is for Elevating. "AI can take it 80% of the way there, but humans should polish and elevate whatever this is before it goes out."

What Pete Is Building

Pete's current projects show what's possible when you stop waiting and start building.

The Family Operating System (FOS). Pete's most ambitious personal project. It captures conversations happening in his home to help him become a better father and spouse, giving him coaching on how to handle situations better. He's a self-described vibe coder now: "I never thought I would ever be coding anything besides HTML and CSS. AI enables me to do things that in my wildest dreams I never thought I'd be able to do."

AI Deep Dive Podcast. A daily, personal podcast built entirely through automation. Claude converts Pete's favorite newsletters into podcast episodes via Notebook LM. He's on episode 156. Find it on Spotify or Apple Podcasts by searching "AI Deep Dive Pete Larkin."

Flight and land agents. Pete has an agent that scans Delta's flash sale inventory daily, looking for international flights out of Salt Lake City under 20,000 points. If one shows up, it sends him an email with the details. He's also running an agent to search for land and property across multiple sources he's given it access to.

Resources: 

TheAIDeepDive: https://podcasts.apple.com/us/podcast/ai-deep-dive/id1836482206  

Utah marketing events: www.theumx.com 

Connect with Pete on LinkedIn: https://www.linkedin.com/in/peteralarkin/ 

Connect with Paxton on LinkedIn: https://www.linkedin.com/in/paxtongray/ 

Looking for an agency that'll be worth the investment? 97th Floor creates custom, audience-first campaigns that drive pipeline and conversions. Get started here: https://97thfloor.com/lets-talk/. 

About Pete Larkin: Pete Larkin began his career in film, running a successful production company before realizing he wanted more influence over the strategy behind the work. That led him into marketing, where his film background and training from one of the nation’s top ad schools helped him build a career as a marketing executive, entrepreneur, and creator.

Today, he has evolved again... this time into GTM engineering and AI orchestration. He now helps organizations and individuals navigate the fast-moving world of AI through agents, workflows, workshops, and consulting.