I spent the last month running an experiment: use AI for every stage of job searching, from finding openings to negotiating offers. Not just ChatGPT-write-my-resume stuff. Custom tools, scripts, actual systems. Some of it worked. Some was a complete waste of time. Here's the breakdown. What Actually Worked 1. Keyword Extraction From Job Descriptions The single most useful thing: pulling keywords from job postings and matching them against your resume before applying. I built a browser-based ATS checker that does this. Paste your resume, paste the job description, get a match score. No server, no data collection. It runs entirely in your browser. The insight: most rejection happens before a human sees your resume. ATS systems filter by keyword density. If the posting says "Kubernetes" five times and your resume says "container orchestration," you're probably filtered out. Dumb? Yes. Fixable? Also yes. Time saved per application: ~8 minutes.…