Find out a remote job is closed to you before you spend an evening on it
Paste a posting and Kairos reads it against your location, citizenship, work authorization and CV, then answers yes, no, or an honest ambiguous — with the line of the posting that decided it. Applications you do send get tracked to an answer.
A senior engineer replies within one business day.
- Paste the posting
- Read against profile
- Verdict + reason
- Tracked in pipeline
Why this exists
Remote job seekers outside the US/EU/Canada waste time on remote jobs that are actually gated by country, timezone, payroll, or work authorization.
Kairos curates work-from-anywhere sources, predicts whether a posting is realistically eligible for a candidate's location, authorization, and CV, then helps track applications through a focused pipeline.
What it does
Everything Kairos handles for you.
Eligibility verdicts
Classifies pasted job text into yes, no, or ambiguous with confidence and reasoning using Heimdall primary and Gemini fallback.
Candidate-aware profile
Uses role, seniority, geography, citizenship, work authorization, skills, links, and active CV context.
Application pipeline
Tracks Saved, Applied, Interviewing, Offer, and Rejected stages with stage history and archive support.
JD-aware writing
Generates personalized application messages from profile, active CV, job description, and templates.
What you get
Concretely, what lands.
Not adjectives — the things that exist afterwards that did not exist before.
- A yes / no / ambiguous verdict with a confidence score and the reason it reached it — not a match percentage with nothing behind it.
- A candidate profile that carries geography, citizenship, work authorization and your active CV, so the verdict is about you.
- An application pipeline with stage history, so you know what you sent, where it went, and when it went quiet.
How it works
From nothing to working.
Build the profile once
Role, seniority, geography, citizenship, work authorization, skills, links and a CV. The eligibility question is unanswerable without them, which is why generic job boards cannot answer it.
Paste a posting
The classifier runs through Heimdall with a Gemini fallback and returns a verdict, a confidence, and the reason — including which line of the posting gated you.
Apply with something specific
Application messages are generated from your profile, your active CV and the job description, rather than from a template with the company name swapped in.
Track it to an answer
Saved, Applied, Interviewing, Offer, Rejected — with stage history and archiving, so the pipeline reflects what actually happened.
Who it's for
Built for these people.
If you are not one of them, we will say so rather than sell you a seat.
- Global-South and non-US remote job seekers in knowledge-work roles
- Candidates who need to avoid country-, timezone-, payroll-, and authorization-gated postings
- Internal proof of applied-AI classification and product build
Why it holds up
What we can show you.
No logos and no usage numbers. These are things that are true about the software itself.
Eligibility AI verdict includes yes/no/ambiguous, confidence, and reason.
Classifier eval infrastructure includes prompt-version tracking, eval run/result tables, and asymmetric precision gates.
Current code gates target yes precision >= 0.95, no precision >= 0.80, and macro precision >= 0.87.
Before you ask
The questions that come up.
The classifier has eval infrastructure with prompt-version tracking and asymmetric precision gates: the current code targets yes-precision at or above 0.95 and no-precision at or above 0.80. It is asymmetric on purpose — wrongly telling you a job is open wastes an evening, wrongly telling you it is closed costs you the job.
Because postings are often genuinely ambiguous, and a system that always answers yes or no on unclear text is guessing. Ambiguous means read it yourself.
It is built for them — the gates hurt most if you are outside the US, EU or Canada. Anyone can use it, but the design assumption is that eligibility is a real question for you rather than a formality.
No. It drafts the message; you send it. An agent that auto-applies at volume is how a candidate gets blacklisted.
Tell us what you'd use it for.
A few lines is enough. A senior engineer reads it and replies within one business day — and says straight away if it isn't a fit.
Or email contact@caffinix.com. We sign a mutual NDA before scoping if you need one.
Built on·Next.js 16 · Postgres/Drizzle · Auth.js v5 · Heimdall · Gemini fallback · Paddle billing
← All products