Dev.to AI 🤖 Ai 👁 0 📖 2 min read

I Tried to Build an AI Agent That Applies to Jobs for Me. The AI Was the Easy Part.

I have been learning Agentic AI for a while now. Concepts made sense, the demos worked, and I felt ready to build something for myself. So I picked a project I actually needed: an agent that applies to jobs for me. Sta

I have been learning Agentic AI for a while now. Concepts made sense, the demos worked, and I felt ready to build something for myself.

So I picked a project I actually needed: an agent that applies to jobs for me.

Stack: Node.js, Playwright, OpenAI.

I thought the LLM part would be the challenge. It was not. The browser part broke me.

What went wrong

  1. Bot detection. Playwright got flagged almost immediately.

  2. Cloudflare CAPTCHA. Once I got past the first wall, this was the second one.

  3. Platform rules. When I finally got through both, I found out the platform does not allow scraping, and my account could get blocked. That is a real risk when the account is the one you use for your actual job search.

What other developers told me

I asked around and got a different answer from everyone:

  • Use paid proxies with proper header params

  • Save the HTML first, then extract the data from it

  • A few more tricks that worked for a day and then stopped

Out of everything I tried, only paid proxies worked reliably.

The real problem: cost

This is a small personal agent, not a product. But look at the bill:

| Paid proxies - Don't want to pay for this
| VPS - Yes, needed in my case
| OpenAI API - Yes
| Other small costs - They add up

And there are already platforms offering this kind of agent at a very low price. So spending every month on my own version stops making sense fast.

What I am thinking now

A few directions I am considering instead of fighting bot detection:

  • Use official sources where they exist: job APIs, RSS feeds, and email alerts instead of scraping pages.

  • Keep a human in the loop: let the agent find, filter and match jobs, and draft the answers, while I do the final click myself.

  • Spend the LLM budget on matching, not on browsing: that is where the agent adds real value anyway.

I have not settled on any of these yet.

My question to you

  • Have you built something like this without paid proxies?

  • Is there a cheaper or smarter approach I am missing?

  • Or is "agent applies for me" simply the wrong problem to solve?

I started this discussion on LinkedIn and some good answers are already coming in. I would love to have your take there too:

👉 https://lnkd.in/p/eCuRfEZB

Or drop a comment here. I read everything.

📰 Read the original article on Dev.to AI

Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.