Somewhere right now, an HVAC company owner is staring at a job board dashboard that says 47 people applied for his service technician opening. He’s paying for AI-powered screening that promised to “surface top talent automatically.” It surfaced three résumés. He interviewed two. Neither showed up for the working interview.

Meanwhile, buried at number 31 on that list is the guy he actually needs. He’ll never see him.

That’s not a hypothetical. It’s the most common story we hear from contractors who call us after their software stopped working for them. So let’s talk plainly about what AI screening does well, where it fails, and why the difference matters more in the trades than almost anywhere else.

What AI screening actually does

Strip away the marketing language and most résumé AI does one thing: pattern matching. It reads the job description, pulls out keywords and credentials, and scores each résumé on how closely the words line up. Years in the title. Certifications spelled the way the software expects. Employment dates without gaps. Clean formatting it can parse.

That’s not worthless. If you’re hiring 200 warehouse associates and every application looks the same, a fast filter has real value.

But a service technician isn’t a warehouse associate. You’re not hiring a set of keywords. You’re hiring someone who will show up on a rooftop at a shopping center, diagnose a package unit that’s been jury-rigged by three previous contractors, explain the fix to a property manager who’s already fielding tenant complaints, and not screw it up. None of that shows up in a keyword count.

Where the software falls down

It penalizes the exact people the trades run on. Career changers, people who learned on the job instead of in a classroom, people who stayed with a small family shop for years and have nothing recognizable on paper to show for it, people who took a year off to take care of a parent. AI reads all of those as risk. Anyone who’s run a crew knows those are often your best people. And it cuts the other way, too: a résumé with five companies in six years can score just fine if the keywords line up. A recruiter sees that pattern and asks the hard question before you waste an interview on it.

It can’t read between the lines. A résumé that says “helped my uncle with his commercial HVAC business for six years” contains no certifications, no formal title, and no company name a database recognizes. A human reads that sentence and hears: this person has been doing the job.

It rewards résumé writers, not workers. The candidate who knows how to stuff a résumé with the right terms scores highest. The tech who’s been too busy running calls to update his résumé since 2019 scores lowest. That’s backwards.

It can’t sit down with the candidate. The single most valuable recruiting tool ever invented is a face-to-face conversation. AI can’t have one. It can’t see that a candidate lights up when he talks about troubleshooting a rooftop unit, or that he’s polite and clear and would represent your company well in front of a facility manager.

It can’t tell you what it missed. When software rejects someone, it doesn’t flag “hey, this one’s borderline, you might want to look.” It just quietly drops them. You never know who was in the pile.

What a human recruiter does instead

A good recruiter starts with a different question. Not “does this résumé match the posting?” but “could this person do this job, and would they stay?”

That means reading every résumé, including the ugly ones. It means talking to people whose applications raise a question instead of deleting them. It means knowing that a two-year gap after a construction job in 2020 probably isn’t a red flag. It means recognizing that “no EPA cert” can be fixed in a weekend while “no work ethic” can’t be fixed at all.

It also means being honest with the client. A human recruiter will tell you when your posting is scaring good people off, when your pay is below market, or when the candidate you love is going to bail in six months. Software tells you what you told it to look for.

A real-world example: the résumé AI overlooks

Here’s a résumé modeled on candidates we see all the time. Names and details are changed, but the pattern is exactly what shows up in our inbox.

A good recruiter reaches out to M. Tillman the same day and sets up a short recorded virtual interview. Fifteen minutes on camera confirms he can talk through a no-cool call on a rooftop unit step by step, that he’s clear and easy to talk to, and that his uncle would happily vouch for him. The recording goes to the hiring manager, who gets to “meet” the candidate on his own schedule: see how he carries himself, how he explains his work, whether he’s someone a customer would trust on their roof, before committing an afternoon to a full interview. The manager watches it that evening. M. Tillman is scheduled for a working interview by the end of the week.

The cost of the résumé you never saw

Contractors usually think about recruiting cost as the fee they pay. The bigger cost is invisible: the tech who would have stayed ten years, the one who got hired by your competitor because their recruiter actually read his résumé, the truck that sat empty for four more months while the software kept surfacing people who never showed, the PM contract you couldn’t take on because you didn’t have the manpower. We’ve seen what daily human screening does to those numbers, and the results are in our case studies.

AI screening is cheap because it does less. That’s the trade. For some roles, it’s a fair trade. For the trades themselves, where the best people rarely look best on paper, it’s not.

Where AI does belong

We’re not anti-technology. We use tools to post jobs, track candidates, and cut down on duplicate work. Software is good at the parts of recruiting that are actually clerical. It’s bad at the parts that require judgment, and judgment is the whole job.

The right setup is a human making the decisions and technology doing the paperwork. Not the other way around.

The bottom line

If your recruiting process would have thrown away M. Tillman, it’s throwing away people like him every week. The fix isn’t better software. It’s a person who knows what a good technician looks like on paper, and more importantly, knows what one looks like when he doesn’t.

Edisto Consulting Group provides fractional recruiting for mechanical and electrical contractors, general contractors, and other skilled trades companies. Every résumé gets read by a person. Every promising candidate gets a recorded virtual interview the hiring manager can watch before deciding who to bring in. One flat monthly fee, no per-hire charges.