Is AI making us better at work or burning us out? ft. Gurmeet Mangat

A solo consulting founder got his first Claude license a week before a coaching session with Gurmeet Mangat. He had never used AI before. In under 45 minutes, with Gurmeet coaching him through his own screen, he set up a workflow that saves him close to a day every week. By 7:00 the next morning, replies to his important emails were already sitting in his drafts folder. His reaction: why did I wait so long?

Gurmeet is back on Tap Into Data for episode 010. He and hosts Jared Westrem and Adam Chartrand talk about why people hesitate to start with AI, how to keep it on track, and how to tell whether it's making work better or just making it faster.

Fear, not skill, is what holds people back

Gurmeet has spent the last few months running workshops, four so far, and he spoke to a room of about 35 CEOs and business leaders. What he hears most is fear. People assume they need to be technical. As he puts it, "You don't need to, but you just got to be curious."

His last slide for the CEOs said to lead with curiosity. He also told them not to try to standardize their AI process yet. In his view, nobody will manage that for the next 24 months, because the models keep changing too quickly.

Adam says the "no time" excuse doesn't hold up. Spend one hour learning to use AI properly, and you can get hours back every day. He has also watched large companies hand everyone a Copilot license with no training, so people just "play around with it." His comparison: a friend's YouTube feed that jumps from 1996 Tiger Woods to spider videos. If you ask AI random questions with no purpose, you get chaotic results.

Treat AI like a new hire, and give it rumble strips

Gurmeet's main suggestion is to think of AI as a person you've hired. It processes information much faster than you do, but its environment still shapes how it behaves. You keep a new employee on track with an SOP or a checklist, and AI needs the same. People worry about "drift" when a new model comes out, but most of them haven't given their AI any boundaries in the first place.

"If there's no rumble strips, no yellow lines in the middle, you're going to drift," he says. "AI is just human. It's just supercharged, but you still need the boundaries."

Jared takes the onboarding idea further. You'd expect to spend a week or two getting a new hire up to speed, so give AI the same time. One method he likes is to let the AI interview you: share some context, then ask it to question you about your work and your business. He's clear about guardrails, though. Giving an agent access to everything and hoping for the best "is not the path for anyone out there."

Gurmeet describes it as graduating an agent through levels of autonomy. At level one, AI handles about 30% of a task and a human handles the other 70%. Over time, that becomes 50/50, then 60/40, then 70/30. Getting to full autonomy will take years.

Read it before you hit send

Adam tells a story about a friend who called a colleague to ask whether he'd written an email with AI. The email made no sense for the conversation it was part of. Adam has seen the same thing in the field: when he and Jared ran drone flights, machine learning returned the data very quickly, but they still had to check that it was accurate before sending it on.

Jared says he's caught himself relying on these tools too much. He noticed his own writing slipping back when Grammarly arrived. Now he gives AI a rough brain dump and polishes what comes back. His fix is to talk through ideas out loud, often by voice while walking the dogs. That makes him explain his thinking, and when he gets a bad answer, he can often see the problem was his question. His rule for drafts and proposals: "Don't take the first response."

Gurmeet calls this using AI as a collaboration partner, not a one-way vending machine. He does the same in his car, asking AI to help him test an idea.

Why smaller companies have the edge right now

Gurmeet thinks small and medium-sized businesses have their best opening in years. They have less IT governance and can move faster than large companies, which tend to be stuck with the tech they already have. Big companies run broad strategy training. What actually paid off in his workshops was tactical training. He asks people what they do every day and to list five things they would hand to a student if they had one tomorrow. That list becomes the set of use cases.

Jared says this is why big companies can spend millions on AI and see little back. Without a strategy, AI ends up as "a glorified research tool." He mentions the daily briefing that lists your emails and meetings, which your phone already does. His advice to people at large companies: if your employer won't train you, learn on your own time so you get your evenings back.

Measure time back, and eye contact

So how do you know AI is working? Adam's test is that his clients stop saying they're "busy." He named three things that save him time: using Boardy.ai, which helps him land 8 to 12 meetings a week; finding RFPs and leads in under five minutes; and recording meetings so an executive summary and action items arrive the same day, not a week later from a note-taker.

Gurmeet's answer is his "eye contact metric." He admits it's an extreme version, but if an AI rollout doesn't have people looking at each other more, it has failed. He doesn't want people turning out 20 hours of work in an eight-hour day and staying glued to their screens. He wants AI to give them more time for coffee, lunch, family, and golf.

To get started, Gurmeet's go-to tool for the past 12 months has been Claude. At 5:00 every day, it checks his CRM, project management tool, accounting platform, and email and writes him a recap. His starter prompt is simple: describe your daily routine to the AI the way you would to a new hire sitting next to you, then finish with, "Now you know what I do. Tell me how you can help me."

The full conversation also covers job fears, home robots, and how soon the keyboard will disappear, so press play to hear it.

Key takeaways

• Fear and the belief that you need to be technical slow AI adoption more than skill does; one coached session saved a solo founder close to a day a week.
• Treat AI like a new hire: onboard it, give it boundaries like an SOP, and raise its autonomy in stages while a human stays in the loop.
• Check AI output before you send it, whether it's an email, a proposal, or drone inspection data, and don't take the first response.
• Small and medium-sized businesses can move faster than large companies, and tactical training built around real daily tasks pays off more than broad rollouts.
• Measure AI success by time given back to people, not just more output squeezed out of each hour.