I Built My Wife a Job Search Automation That Texts Her Gigs Before Coffee

A single text message put a job in my wife’s pocket before her competition even knew it existed — and the job search automation that sent it (I call it the Job Sniping Engine) is the thing I rebuild from scratch in this episode, live, on camera, right down to the part where it blew up in my face.

My wife Melanie is a traveling dietitian. Every morning she burned real time refreshing the same job board, hunting for her next contract. So her “homework” for me was simple: stop her from refreshing a website like a maniac all day. This is how I built her a job search automation that does the hunting for her.

In a contract job market, speed wins

Traveling dietitians fight over the same gigs on one board. A job posts, the first qualified person to apply gets the call, and the good ones are gone in hours — sometimes minutes. The whole goal was a robot that watches the board 24/7 and texts Melanie the second a gig in one of her states goes live. No refreshing. No missing the good ones. Just a buzz in her pocket before her competition has opened a laptop.

The map: what this job search automation is made of

Before any code, here is the whole stack on one screen and how the data flows:

• WordPress signup page (hosted on GoDaddy) — the front door where a dietitian picks her states.
• Go High Level — the CRM where every person and their job history lives.
• n8n — the brain that orchestrates the whole thing every morning.
• ScrapingBee — grabs the live listings off the job board.
• Gmail + SMS — the alerts that land on her phone and in her inbox.

A dietitian signs up once on our signup page, the data flows into the CRM, n8n runs the routine on a schedule, and a text plus an email go out. Sixty seconds, top to bottom.

The front door (the easy win)

First, the signup: a clean form where someone picks the states they want and drops in their number. They sign up, a contact lands in the CRM, and a verification email goes out. I built it in WordPress, and honestly Claude wrote nearly all of it in one block of code while I sat there and talked to it. Small, visible win — momentum before the hard part.

The brain

Next, the logic that makes the job search automation actually smart. An automation pulls every contact, figures out which states people actually care about, and builds the watch list from the data instead of me hardcoding it. Contacts in, the exact set of states to monitor out. Add a new person in a new state tomorrow and the robot just starts watching that state too. Nobody touches code.

Where the job search automation broke (the part that always fights me)

Here is the real struggle. To grab live jobs, the bot has to read a website that does not want a robot reading it — and three things broke at once, on camera.

First, the page is JavaScript-rendered, so a normal grab comes back empty because the jobs load after the page does. You need a tool that renders the page like a real browser before it reads it. Second, the classic gotcha: the page hands back three paragraph tags per job — empty, real, empty — so the raw output is garbage. Third, the site rate-limits the bot and starts throwing errors mid-run.

I left it ugly on purpose. Watching me sit in the confusion — that is wrong, that is empty, where is the actual job — is the honest part. That is web scraping in the real world, not the tutorial version.

The fix

Two fixes, both dead simple once you see them. One: a five-second wait inside the loop, so the site stops treating the bot like an attack and quits rate-limiting it. Two: grab the right paragraph — there are three per job and I only want the middle one, so I count by threes and pull the one in the middle. Run it again, and there it is. Clean jobs, every time.

The payoff

Hit run. The digest builds, person by person, state by state. Then the money shot: a real phone buzzes on the desk with a real job alert. That text at the very start of the episode? That is the Job Sniping Engine working. It works.

Build your own job search automation

The best part is none of this is exotic. The same pattern — watch a source, clean the data, route it to the people who care — is a template you can steal. Swap the job board for any site that posts time-sensitive listings, and you have the same job search automation pointed at a new target. That is the whole point of Extreme AI Makeover: a real build for one person that turns out to be a tool for anyone.

What is next

I cut the episode right where it got loud, because the full meltdown — me losing my mind over a two-word state bug for three hours — earns its own video. That blooper reel is dropping next. Subscribe and hit the bell so you do not miss me come apart in real time.

And if you run a business with a real workflow problem that is quietly driving you nuts, become a Build Partner: show up for three episodes, be honest about what is working and what is not, and actually implement what we build. Email eam@x7ai.ai.

Real builds. Real struggle. Real results.