If you’ve decided it’s time for your team to go all in on AI, it’s important to approach this technology with a clear understanding of its capabilities, where guardrails are needed and how to message your usage to customers.
Avoid Creating AI Slop
One of the major complaints against AI is the belief that all it produces is slop, but Mike Lysecki, the CEO of LeanScaper, argues most people formed this opinion on their first try a few years ago.
“The models have moved fast, even in the last six months – what used to come out as generic mush now comes out genuinely good, often better than what a rushed person would produce,” Lysecki says. “And let’s be honest: human slop is real too. We’ve all gotten a careless estimate from an actual person. Slop isn’t unique to AI; it’s a symptom of low effort, wherever it comes from.”
Opting to use AI in your business isn’t a license to mentally check out when it comes to work. Lysecki says whoever uses that AI tool must own the result.
“It doesn’t matter whether AI drafted it or built the whole first pass – if it goes out with your name on it, you own it, not the tool,” Lysecki says. “That single mindset is what protects your brand, because it puts accountability where it belongs: on the person.”
Lysecki says what AI produces really comes down to proper prompting and curating.
“Spend five careless minutes prompting, and you’ll get a careless result, the same as if you’d rushed it yourself,” Lysecki says.
Yet when you take the time to teach AI your standards, it can perform to it consistently. You should also ensure your team knows what good looks like and what you’d never put your brand name on.
“The companies that lose their edge here are the ones who publish AI’s first draft unexamined,” Lysecki says. “The ones who get stronger teach their people to be demanding editors.”
Set Guardrails
When rolling out AI, Lysecki says the best rule of thumb is to treat it like a brand-new employee.
“No matter how talented, on day one it doesn’t know your company – your standards, your process, what ‘great’ looks like around here,” Lysecki says. “You wouldn’t let a first-day hire send work to your best client unsupervised, and AI is no different. You look at the work before it goes out, and you keep looking until it’s earned your trust.”
Nothing should go out without a human set of eyes reviewing something AI produced. However, you shouldn’t just clean up the output. Fix the instructions and prompt that produced the result.
“Correcting the output solves today’s problem once; correcting the instruction solves it permanently,” Lysecki says. “That’s the real difference between AI and a person: with someone new, you often give the same correction again and again. With AI, you fix the instruction once, and it holds. You’re not just reviewing the work – you’re upgrading the system that produces the work.”
Lysecki says while verification will be the heaviest in the beginning, as your guardrails are built out, your reviews will shift from fixing to spot checking.
You also have to consider your access and permissions as AI is only as safe as the guardrails around it.
“You have to be deliberate about how it exposes information – making sure people only see what their role should see,” Lysecki says. “Your foreman doesn’t need visibility into payroll; a seasonal crew member shouldn’t be able to pull up customer contracts. If you roll out AI without controlling access by role and security permission, you can accidentally make sensitive information a lot easier to reach than it ever was in a filing cabinet.”
Who Has Final Say?
Lysecki stresses that while AI can make your people faster and sharper, humans still need to be held accountable anytime money moves, the record of truth changes, safety is on the line or a relationship is at stake.
While having AI read your financials is low risk, you should always have a person on hand for anything that changes the record of truth. Lysecki cautions against letting AI enter or edit data unsupervised.
When using autonomous equipment in the field, a crew’s safety should still be a person’s responsibility at the end of the day.
Similarly, when navigating a legal or safety issue, firing someone or bidding on an unusual job, while AI can inform decisions, it shouldn’t make the final call.
“When you’re dealing with a real complaint, a damaged property, a billing dispute, or an unhappy long-term client, that conversation needs a human being who can read the room and own the relationship,” Lysecki says. “Let AI prepare you; don’t let it replace you at the moments that actually test the relationship.”
Be Transparent with Clients
Lastly, it’s critical to be transparent with your customer base about how you use AI. How you communicate this to clients will impact how they view your business’s usage of it.
“Just don’t present it like a disclaimer, some fine print you’re obligated to confess,” Lysecki says. “That framing signals that you think it’s something to be nervous about – and if you treat it like a liability, your client will too.”
If you talk about AI as an asset and confidently share how this investment will allow you to serve them better, it serves as a selling point, rather than an advanced apology. For instance, ‘We capture all the details of your project up front and run them through our system so nothing gets missed, your crew shows up knowing exactly what your job needs, and the work gets done right the first time,’ sounds far more assured than a sheepish ‘Just so you know, we use AI.’
“No homeowner or property manager loses sleep over which tools you use – they care whether the job got done right, whether anything slipped through the cracks, whether you were easy to work with,” Lysecki says. “Frame AI around those outcomes and transparency becomes an advantage: you come across as a modern company investing in doing better work, not one hiding something or apologizing for it.”
