How Most People Use AI for Job Hunting Is Dead Wrong

Open ChatGPT. Paste the job description. Say “polish my resume.” Copy, paste, submit. Then… silence.

What’s the problem? Single-shot generation. You’re asking one AI to be both writer and judge in a single conversation. What you get back is lukewarm mush — grammatically correct, zero edge, the kind of resume an ATS buries on sight.

Job hunting is subjective, multi-step, and emotionally brutal. Most people fire off dozens of applications and land a handful of interviews. That’s not a talent problem. It’s a process that was never engineered.

A Laid-Off Geophysicist Turned Job Hunting into an Assembly Line

According to the source post (none of these figures independently verified): there’s an open-source GitHub project called ai-job-search with 29k+ stars. The author, Mads Lorentzen, was a geophysicist. After getting laid off, he built a job-hunt Agent framework with Claude Code, sent 69 applications, landed 20 first-round interviews, and successfully pivoted into an AI engineer role.

69 applications, 20 interviews. That’s close to a 30% first-round conversion rate. The average for spray-and-pray applying? Usually under 5%. Per the source, he published this funnel data on LinkedIn himself.

But the numbers aren’t the point. The point is the method: he decomposed job hunting from “voodoo” into a standardized workflow that AI can execute, verify, and iterate on.

The Design Actually Worth Stealing: Dual-Agent Review

Per the source post, the core mechanism is a Drafter + Reviewer architecture:

• Drafter: reads your career profile, generates a tailored resume and cover letter for a specific role.
• Reviewer: a separate Agent launched in a fresh context, uncontaminated by the draft. It researches the target company on its own, then tears the draft apart from a third-party perspective.

This design hits the exact weak spot of single-shot generation. Agent one writes. Agent two attacks with fresh eyes: “This experience is irrelevant to the role.” “That verb is weak.” “The company just did layoffs and your cover letter doesn’t address it.” Then it sends the draft back for a rewrite.

Per the source, the whole flow runs on four commands: /setup builds your career profile (imports PDF resumes, LinkedIn exports, or runs an interview-style intake), /scrape finds and ranks openings (ships with Danish boards like Jobindex, Jobnet, Jobbank — extensible to any region via commands), /apply generates application materials in one shot (match scoring, LaTeX formatting, ATS keyword validation, PDF page checks), /interview produces an interview prep pack (company research, STAR story mapping, mock interviews).

All data stays local. No SaaS platform touches it.

What This Means If You’re Here to Make Money

Two paths.

Path one: use it yourself. If you’re job hunting — especially in tech — stop blasting one generic resume everywhere. Fork the project (github.com/MadsLorentzen/ai-job-search), or rebuild the logic yourself with Claude: profile database + job parsing + dual-Agent review + ATS validation. Even if you only implement the “Reviewer critique” step, your resume will outclass 90% of the competition.

Path two: sell it as a service. This is what I actually want to talk about.

Step 1: Package the Workflow as a “Resume Sniper” Service

Search “resume review” on Fiverr or Etsy and you’ll find a sea of sellers at $15–60, all doing manual edits or single-shot GPT output. You’re holding dual-Agent review plus ATS keyword validation. You’re not even playing the same game.

Pricing: Basic at $29 (resume optimization + ATS keyword report). Pro at $89 (resume + cover letter + target company research). Premium at $179 (adds interview prep pack + mock Q&A bank). Your cost? A few cents of API calls per run.

Step 2: Acquire Clients with Before/After Content

Post on Lemon8 and Pinterest: left side, the client’s original resume ATS parse (31% keyword coverage). Right side, your optimized version (87% coverage). That visual contrast is the strongest conversion asset in the resume business. Then cut a TikTok showing “watch an AI reviewer savage this resume line by line.” Conflict sells.

Peak seasons: September–October and January–February hiring waves. Seed your content a month early.

Step 3: Upsell from Resume to Full Job-Hunt Management

Once a client lands interviews, upsell interview coaching. Above that, sell “job-hunt retainer” packages to career switchers and international grads at $200–450/month. Weekly deliverables: curated job lists, tailored application materials, interview debriefs. Ten retainer clients is roughly $3,000/month — and Agents already handle 80% of the grunt work.

If your English is solid, list directly on Fiverr and Upwork. “ATS-optimized resume with AI review” goes for $50–150 a pop, and the competition is far less cutthroat than you’d expect.

The Bigger Lesson

The real value of this project isn’t job hunting. It’s the demonstration that any service business that is “highly subjective, multi-step, with verifiable outcomes” can be rebuilt as an Agent workflow and delivered at 10x efficiency.

College application essays. Grant proposals. Immigration paperwork. Compliance document review. Same shape, all of them. Find the step where “clients pay a premium but delivery is painfully tedious,” engineer it with the Drafter + Reviewer structure, and you own an asymmetric weapon: competitors burn headcount, you burn API tokens.

Do one thing tonight: pick a paid service category you know well, write out its full delivery process, and mark which steps an AI can draft and which step a second AI should critique. That sheet of paper is the blueprint for your next business.