A daily system that finds, filters, and builds the application before I open my laptop

What It Is

This system runs once a day without me. Before I've opened my laptop, it has already swept LinkedIn hiring posts and a fixed list of job boards, checked every candidate against a written filter (location, a pay floor, how fresh the post is), and for anything that clears those gates, built the entire application package: company research, a positioning strategy, a tailored resume, a cover letter, and outreach drafts for email and LinkedIn. My part is a short daily review, then deciding what to send.

Why I Built It

A review of my own application history showed a clear pattern. A large share of the companies I'd successfully engaged came from a hiring manager or founder posting "we're hiring" directly to their own LinkedIn feed, not a job board listing. Those posts move fast, comment sections fill with competing applicants within hours, and by the time I'd notice one on a normal scroll through LinkedIn, the window was often already closing. I needed something watching for that pattern every day, not me remembering to check.

The second half came from a different problem. Once a role qualifies, building the application, researching the company, developing a positioning argument, tailoring a resume, and writing a cover letter that isn't generic, takes time. Doing that well for every lead that came in would have meant either working through a backlog every night or lowering my bar for what counted as worth applying to. I wanted the bar to stay high without the time cost rising with it.

What It Does

Every run checks three sources against each other first, so nothing gets logged twice: my own application history, my LinkedIn applied history, and a running candidate log. Then it searches a fixed list of job boards and sweeps LinkedIn for hiring posts and live listings. It reads the posting text itself rather than trusting a search snippet or an aggregator's tags. A listing site's "still open" or "remote" label turns out to be wrong often enough that I stopped trusting it directly. Every candidate that survives gets checked against hard gates, geography, a pay floor, how recent the post is, before it's scored against a written fit filter.

Anything that clears both stages gets the full build: the job details and positioning notes, research on the company's product, market, and hiring signals, a strategy hypothesis, a resume tailored to that role with a rationale for the choices made, a cover letter, and outreach drafts with contacts found through real searches rather than guessed. Where a question needs something only I can answer honestly, a rate, a personal story, an unverified contact, the system stops and flags it instead of filling in a guess. I get one notification a day with what's ready, review it, and decide what goes out.

What It Taught Me

Automating the finding was the easy part. Building in enough verification was harder. One posting's own copy read "work from wherever you're most productive," but that line said nothing about the real location rules buried in the company's actual application system. I'd have promoted that role on the strength of a sentence that turned out to be marketing copy, not a rule. Checking each posting's location claim against the employer's own page took more care to build than the daily sweep itself.