It starts empty
One card on the whole screen: create the person you are hunting for. Nothing else is shown, because nothing else matters yet.
AI job-hunt agents · every IT role
The job hunt is brutal right now. Let the agents do the grind.
Qiksy Offer runs Claude agents that sweep the whole market (company sites, ATS, boards, hiring threads and early signals), match roles to your profile, draft tailored applications, and keep watching so new roles come to you. Every IT profession: developers, QA, PMs, designers, DevOps.
Start your job huntWatch it work on sample dataFree to start · runs on your own Claude · nothing leaves your machine
1,240+
sources swept every search
dev · QA · PM · design
every role in IT
cover letters
tailored, from your resume
100% local
your Claude, your machine
What the agent is doing while you wait
First finds land in three to eight minutes, a full sweep takes about fifteen. This is why: every posting is opened and read in full, not skimmed from its headline. You can close the tab; the run keeps going on your own machine.
The wording in the log is the tool’s own. The run is redrawn here so you can see what a search looks like without having to start one.
The hard part, done for you
Job hunting is hours of tab-juggling and copy-paste. Qiksy Offer turns it into a queue of agents.
Doing it by hand
With Qiksy Offer
Refreshing ten job boards, missing the fresh ones
Agents sweep every board, ATS and hiring thread at once, and nothing slips past
Re-reading each posting to guess if it fits
Every hit lands scored against your profile, with why-it-fits notes
Rewriting the same cover letter for the hundredth time
A tailored letter drafted from your resume for each role, and you just review
Filling the same application form again and again
The agent fills and can submit the form for you, while you keep control
This is what it actually looks like
Taken from the working tool on a live database, not from a mockup. Every card: the role, a match score, the agent’s reasoning against that specific résumé, the region, the conditions and the pay scale.
IT & software
Stack, level, work format, English. The agent reads the posting itself rather than the board’s tags, so “React” in a skills list and “React” in the daily work are two different things to it.
Healthcare & care work
Nurses, carers, care assistants. Here the stack decides nothing; what decides is what the ad does not show: language level, diploma recognition, shifts, employer housing, the pay scale. The agent checks exactly that, and says so when the ad stays silent about recognition.
For offline professions the hunt does not run “across Germany”. It runs across specific states, in the words that profession is called by locally.
Two industries are open. Yours can be next.
An industry is not a label here. It is what the agent knows: which sources to sweep, what the profession is called locally, and what actually decides the hire in that market.
Open now
IT & software
Every role in the field: developers, QA, PMs, designers, DevOps, data. Career pages, ATS, boards and hiring threads; remote scope and seniority read from the ad itself.
Open now
Healthcare & care
Nurses, care assistants, live-in carers. Searched in the local language under the local job title, and scored on what the offer really turns on: diploma recognition, visa, language level, housing, pay scale.
On request
Your industry, built to order
Logistics, construction, hospitality, accounting, trades, whatever you hire for. The interface stays this one; what we assemble behind it is your engine: your sources, your job titles, your fields, your scoring.
Ask for your industryBuilt to order, one client at a time, not a checkbox in a mass-market plan.
How it works
The whole flow, screen by screen, in the order you meet it. Every frame is the tool itself, recorded rather than drawn.
The profile
who you are hunting for
Tell it about yourself
Speak or type, in your own words. No twenty-field form: the agent asks for a story and turns it into a résumé.
Claude assembles the résumé
Contact, headline, experience, skills, all filled in from what you just said. Every field stays yours to edit.
Candidates & roles
what they can actually get
…and names what is missing
Country of residence, languages, notice period. It says the gaps out loud instead of leaving them quietly blank, because each one changes what the search can find.
A role belongs to a person
Roles branch off the candidate, taken from their résumé. Plus adds one, minus takes it out of the hunt without deleting anything.
More than one person
Add a second candidate and the rail groups by owner: a partner, a team, or an agency’s whole bench on one board.
The hunt
agents sweep the market
How fresh, and go
Pick the window (3, 7, 14 or 30 days) and one button runs every role at once.
The agents work
Radar, live log, progress. You can close the tab: the run keeps going on your own machine, on your own Claude.
The board fills up
Findings land one at a time, best match first. Tiles or rows, your call.
Triage & apply
from a board to a sent application
One card, read close
The score, why it fits you, the terms. Under every card: pin it, leave a note, or archive it, and archived is never gone.
Filter when it gets busy
Status tabs across the top, then candidate, geography, freshness and a text filter over everything.
The letter for THIS ad
Open a card and the agent drafts against that posting’s own wording and requirements. It fills the form and stops before sending, so you press send.
Your data never leaves your computer
Candidates, résumés, everything the agents find and every letter they draft are written to one file on your own disk. There is no account, no sync, and no cloud of ours holding a copy. We could not read your data if we wanted to.
One file, on your machine
The whole database is a file in your own home folder. Nothing is uploaded anywhere: the agents run on your computer, on your own Claude sign-in.
Moving to another computer
Settings → Your data → Export saves that file. On the new machine, Import brings it back, either merged with what is already there or replacing it. You carry the file; it never passes through us.
Nothing to leak, nothing to delete
We hold no copy, so there is no account of yours to breach and no “delete my data” form to fill in. Removing the tool removes the data with it.
The one consequence to know: since we keep no copy, a new computer starts empty until you bring the file across. Your licence key is not in that file; it stays in your purchase e-mail.
Pricing
The full search is the same on every plan; you pay for automation and scale.
Free
Try it on your own hunt
$0
- The first week is fully open: several candidates, every Pro feature
- 1 candidate, every title they fit, and the full search on you
- Searches everywhere: your Claude sets the pace
Pro
For a serious job seeker
$129/yr
Everything in Free, plus:
- Monitoring: new roles surface on their own, before the market
- Auto-applications with tailored letters, written for you
- Hiring leads: funding rounds & early signals
Team
Recruiters & agencies
$99/seat/yr
Everything in Pro, per person:
- One more candidate per seat
- A separate pipeline for each person
- From 2 seats · yearly, no auto-renewal
one price worldwide · yearly, no auto-renewal · runs on your own Claude
Not this engine, but yours
Your industry is not a checkbox. It gets an engine of its own.
The interface stays this one. What sits behind it is a different machine: your sources, your job titles in the language of your market, your fields on the card, and your own scale of what counts against a match.
We have already done it once, and you are looking at the result
IT & software
Career pages and ATS (Greenhouse, Lever, Ashby), plus global boards and hiring threads. Twenty-six roles in one vocabulary, scored on stack, seniority and how real the «remote» in the ad actually is.
Healthcare & care
None of the above. Public employment services in 25 countries, sector boards in 27, large care operators in 13, diploma-recognition programmes in 14. Searched as Pflegefachkraft, pielęgniarka, infermiere (117 titles localised into 25 languages) and scored on recognition, language level, visa, employer housing and shift pattern: twelve fields an IT posting simply does not have.
Two industries, one interface, two different machines underneath, down to the scoring, which shares not a single coefficient between them. That is what «one more industry» really costs, and we have paid it once already. It is an assembly, not a switch in the settings.
What gets assembled for you
Sources, named one by one
Where the work in your field actually gets posted: employment services, sector boards, the employers themselves, agencies, specialist programmes. Not «we search the internet», but a list you can read and correct.
The profession vocabulary
Your job titles, grouped the way your field groups them, and translated into the language of every market you hunt in, because the English term leads to aggregators, not to vacancies.
The fields on the card
What the agent is required to find out about every opening: the things your decision actually turns on. In care that is diploma recognition and housing; yours will be your own, and we write them down at the discovery.
The scoring
Your match scale and your deductions: what lowers a score and by how much. Every deduction is shown as a line a human can argue with («−25 · on-site only») instead of hiding inside a final number.
Filters and geography
The cuts your market is really sorted by: countries, administrative regions under their local names, freshness, statuses, whatever you filter on when you work.
What comes out
Letters written from the résumé, application forms filled in your own browser up to the send button, monitoring that watches for new openings, and an export of the whole database.
Why an engine built for one field beats a general-purpose one, and where it does not
A general-purpose tool has to fit everyone. That is its condition, not its flaw: it searches where searching is the same for everyone, asks what makes sense for everyone, and scores on what everyone has. An engine built for one field carries no such condition. It goes to the addresses that are real in that field and names them one by one. It searches with the word the profession is called by locally, not with a translation from English. And it finds out what the deal actually turns on there, which a general tool has nowhere to put, because it has no field for it. What we do not claim: that we will find what is not in open sources, or that another tool will find less. What we do claim is that where your field has to be looked at, ours looks, because it is written down rather than guessed.
How this actually works. We start with a paid discovery: your field, your sources and the signals you select on, written down, and the quote comes from that document, so you see the price and the timeline before the work starts, not after. The build then goes module by module at a fixed price each, not by the hour: anything not in the discovery is a separate line, not a surprise. After that, support under contract, because sources change and someone has to watch them. Inference runs on your own Claude subscription: there is no usage bill from us. What we will not do is name a timeline before the discovery. And if your field sits close to one already open, we will say so, because opening it as a vertical inside this tool is cheaper than building a separate system.
Discuss a buildMore from Qiksy Studio
Two more doors while you are here
Let the agents do the hunting
Add a candidate, run it on the Claude already on your machine, and the market comes to you: matched roles, ready letters, one clean pipeline that keeps watching for new openings. First candidate is free.
Start your job huntFree to start · nothing leaves your browser · runs on your Claude sign-in
Follow the build
We're building Qiksy in the open: new tools, real progress, the occasional stumble. Get the short note when something ships.