Case study · AI text screening · 2026Konrad Kopczynski

Hiring hourly workers faster: an AI that screens applicants by text within minutes

In home services you compete for workers on speed as much as you compete for customers. I built a text screener that answers every applicant within minutes, covers what the hiring manager needs to know, and gets good people to an interview the next morning. Cost per hire went from about $1,000 to $100–150.

MinutesFrom an application to the first text, any time of day

$100–150Cost per hire, down from about $1,000

Next morningWhen a good applicant from 8pm can be in an interview

3Traits scored for the manager: friendliness, hustle, quality

Watch it run: 3.5 minutes

The real hiring flow, with applicant names, phone numbers and the company's branding blurred.

The problem

Someone applies on a Thursday at 8pm, after their shift. They're applying to five or ten places that night. They want to know what they can start on tomorrow.

The usual process is to read a resume that says little about the job, then set up an interview. By the time anyone calls back it's been a week, and they've taken another job. That doesn't mean they'd have been a bad hire. It means someone else answered first.

Before this, each technician hire cost us about $1,000.

How it works

A careers page per location. Built from one settings screen: the brand, pay range, weekly earnings, schedule, open spots, where the crew meets, the career path and the questions to ask. It leads with what someone makes per week, not per hour.
Applying takes seconds. A name and a phone number, or a text to the hiring line.
A text back within minutes. The screener says up front that it's automated. Before it did, people kept asking "are you an AI?". Once it says so, they don't mind.
The same topics every time. Where they live and whether the commute works, a valid driver's license, the schedule, work authorization and hands-on experience. It asks follow-ups like a person would. The license question goes out word for word every time, so we can search for who confirmed it.
Plain rules for the knockouts. Hard requirements are rules, not AI judgment.
A scorecard for the hiring manager. Friendliness, hustle and quality, each with a star rating and the reason, plus red flags, a summary and the full conversation. It goes to the team inbox, or by email. The full topic list and rules are on their own page.
A real interview offer. The manager texts back from the same line and books an interview, often for the next morning at the crew's meetup spot. Some people start the day after.
Follow-ups. Anyone who stops halfway gets a nudge after 48 hours, and a separate check catches referrals.

Where people stay in charge

The AI scores each screen and flags concerns, but it never turns anyone down. The manager reads the full conversation and makes every hire.

Each flow runs on a scheduler with the settings in one place, so a new location gets its own page, number and questions without new code.

What it's built on

A text line from Quo, a Claude model that gets each message and answers with the whole conversation in front of it, workflows on a scheduler, and the team inbox in Missive. It's one piece of Titan, the operating system I built and rolled out to 120+ home services locations.

Try it yourself (845) 303-2097 Text the hiring line and you'll get the screener. It's a live hiring line, so a person will see your replies.

If you're trying to do this in your operation, here's how I work with teams fractionally, or just email me to compare notes.