AI job search
An AI job search
that never applies for you.
Two jobs, and we do exactly two. Watch the market so you hear about a posting on the day it appears, and write a CV for the one posting you decide is worth it. Everything a product in this category usually promises after that, we refuse on purpose.
Counted on 21 August 2026
What is behind it.
- 667,633
- postings storedLive in the corpus right now. Every cycle adds more, and nothing is thrown away when it closes.
- 514,893
- with the full description77% of the corpus, so a filter can read the body of the ad and not only its title.
- 30 min
- on a board somebody watchesEverything else is swept daily. The scheduler ticks every 15 minutes.
- 34
- board systems parsedGreenhouse, Lever, Ashby, Workday, SmartRecruiters, iCIMS and the rest, plus the big aggregators.
How it works
Store the market, then search it. Not the other way round.
One scanner, everybody's market
There is no crawler per customer. A single sweep covers every board we track: the boards somebody is actively watching every 30 minutes, the rest once a day as a statistic. You signing up costs the engine nothing, which is why watching is free and stays free.
Judgment happens after collection
Nothing is filtered at collection time, so nothing is discarded before you have said what you care about. Your criteria are a query over everything stored, which is why narrowing a rule tonight narrows last month too, and why a brand-new account searches a market that was collected before it existed.
The hour, not the week
A posting is read hardest on its first day and is effectively closed once it is a hundred applications deep. A match reaches you by email or Telegram with the posting attached and the reason it matched, and then it waits for you.
A CV for the one you pick
Approve a posting and a document is written against it: your evidence, selected and re-stated in that job's own language, reviewed by 25 independent readers, checked again as a rendered PDF. About an hour, measured at 34 to 45 minutes.
The line we do not cross
We will not apply
for you. Ever.
This is the first thing people ask, so here is the whole argument rather than a policy line.
An application is a statement made in your name. A bot filling an employer's form is you saying things you did not read, to a company you did not look at, under a signature that is legally yours. When it gets a fact wrong, it is your name on the wrong fact, and the recruiter's only reasonable conclusion is that you sent it.
There is a second reason, and it is less noble. Mass submission is what makes a posting three hundred applications deep by Tuesday. That pile is the problem we sell into. Building a machine that makes the pile taller and then selling you a way through it would be a strange business to be in, and we would deserve everything that came of it.
So we do the two things a machine is genuinely good at, and stop. Pressing send stays yours.
Alerts
How you actually hear about a job.
Matches arrive by email or Telegram, whichever you turn on, and each one carries the posting and the rule that matched it so you can tell in five seconds whether to open it. The sweep runs around the clock on its own cadence; how often you are told is a per-track setting, because a signals track you check weekly and a search you are running hard this month should not sound the same.
What the scanner does all night
Boundaries
What this is not.
Not an auto-apply bot
No submissions, no forms filled, no messages sent as you. See above; it is the whole position.
Not a job board
We do not host jobs or sell placements to employers. The corpus is collected from public career boards, and no employer pays us to be in it.
Not a resume builder
There is no editor to arrange your own bullets in. You write a profile once and every CV is generated from it against a specific posting.
Not the whole world
US tech: engineering, product, data and the leadership roles above them. Other regions come later, and we would rather write that here than let you find out after signing up.
Go deeper
The two halves, separately.
The resume writing service
What you get for a payment, how it differs from a human writer, and what the pipeline does to a draft before you see it.
Read the service page02Tailoring, as a method
How a CV is actually bent toward one posting, worked line by line, whether or not you buy anything.
Read the method03The keyword finder
Paste a job description and see what it demands, what it merely prefers, and which of its terms your CV never mentions.
Open the toolFair questions about an AI job search.
Does an AI job search actually work?
The matching half does, because it is a search problem: a machine reading 667,633 stored postings against your criteria every night beats you refreshing four career pages. The writing half works when the profile behind it is good. Neither half gets you hired, and a product in this category that talks about outcomes rather than mechanics is selling you the part it does not control.
Do you apply to jobs for me?
No, and we will not build it. An application is a statement in your name, and mass submission is what makes the pile we sell into. Both reasons are on this page in full.
Is it free?
The account, the market watching, the criteria and the alerts cost nothing, and that is not a trial. Money changes hands the first time you decide a posting deserves a CV, once, for that CV.
Which job sites do you scan?
Company career boards on 34 different board systems, including Greenhouse, Lever, Ashby, Workday, SmartRecruiters, Workable and iCIMS, plus the large aggregators. Boards are added when somebody wants a company that is not in yet, and adding one costs you nothing because one sweep serves everybody.
How quickly do I hear about a new posting?
The scheduler ticks every 15 minutes. A board somebody is actively watching is fetched at least every 30 minutes; every other board is swept daily as a statistic. So a job on a company you track is usually in front of you within the hour it goes up.
Do I have to tell you which companies I want?
No. You can name the companies you would actually join, or search everything stored and let the engine tell you who is hiring. Naming a company does not slow anything down, because its board is already in the sweep.
Does it work outside the US?
Not yet. The corpus is US tech. Postings elsewhere exist in it as an accident of collection rather than as coverage we would defend, and we would rather say so than sell you a search over a market we do not really have.
Is my profile used to train models?
It is used to write your documents. It is not sold, not handed to employers, and not turned into training material for public models. The privacy page is the long version and it is short.
Describe the job.
The engine starts tonight.
Free account, free watching, one payment when a posting is worth a document. Sign in with Google, a passkey, or an email address.
Start free