How it works
Most large employers run applications through an AI scoring system (ATS), and a person reads only the few top-ranked CVs / resumes. AI Job Hunter evens the odds for the applicant. It analyses a job description the way the ATS does, so you can tailor your CV to the keywords that put you among the top few.
AI Job Hunter finds, each with the evidence sentence from the job description:
- the knockout requirements that reject an application automatically
- the Required and Preferred keywords your CV is scored on
- the main duties and notes
- how the ATS scores each keyword, and how to get the top score: where to place it on your CV, in what exact form, and why recent experience counts more
You know before you apply whether you qualify, and which exact terms your CV / resume needs to make you a top candidate. This is free.
Why not just ask an AI model yourself?
We tested it and gave the same five job postings to a leading business LLM. We asked it for the ATS keywords a candidate should include in their CV, then ran the same postings through AI Job Hunter.
- The LLM listed 130–149 keywords per posting with no structure. It did not flag any of the knockout requirements as pass/fail. It listed duties and company description as if they were requirements, and a third of its keywords are not in the posting at all. You cannot tailor a CV to a list like that.
- AI Job Hunter returned 19–43 items per posting, with all knockouts listed first and each keyword shown with the evidence sentence from the job description.
| LLM / AI-JH comparison, 5 test JDs:LLM / AI-JH | CoStar | Walmart | Tiffany | NSI | Intuit |
|---|---|---|---|---|---|
| Knockout requirements identified | 0 / 4 | 0 / 1 | 0 / 1 | 1 / 3 | 0 / 2 |
| Required / Top keywords identified | 1 / 7 | 0 / 7 | 2 / 10 | 0 / 7 | 6 / 33 |
| Required + Preferred keywords found | 14 / 22 | 10 / 26 | 11 / 17 | 6 / 12 | 18 / 38 |
| Noise (duties presented wrongly as R/P) | 16 / 0 | 16 / 0 | 17 / 1 | 23 / 0 | 19 / 0 |
| Noise (company description wrongly as keywords) | 20 / 0 | 13 / 0 | 0 / 0 | 2 / 0 | 5 / 0 |
| Not in JD (invented and added) | 36 / 0 | 60 / 0 | 77 / 0 | 37 / 0 | 23 / 0 |
| Total keywords listed for CV tailoring | 137 / 29 | 149 / 30 | 139 / 21 | 147 / 19 | 130 / 43 |
Each cell: LLM / AI Job Hunter. Test of 20 October 2026; AI models change over time.
What the rows mean
- Knockout requirements identified
- Pass/fail items in the posting, such as a degree, years of experience, a licence, a security clearance or work authorisation. For the LLM, the count is knockouts it mentions at all.
- Required / Top keywords identified
- The job posting’s Required keywords that appear in the LLM’s own list of top keywords to prioritise / all of AI Job Hunter’s Required keywords.
- Required + Preferred keywords found
- The job posting’s Required and Preferred keywords that the LLM also lists / all of AI Job Hunter’s Required and Preferred keywords.
- Noise (duties presented wrongly as R/P)
- Terms taken from the duties or responsibilities lines but presented as requirement keywords.
- Noise (company description wrongly as keywords)
- Terms taken from the company description or the benefits section and presented as keywords.
- Not in JD (invented and added)
- Keywords that no line of the posting states, even in other words (synonyms and rephrasings count as stated).
- Total keywords listed for CV tailoring
- LLM: every distinct keyword in its reply. AI Job Hunter: knockouts, Required and Preferred keywords, and main duties.
What it does today, and what’s coming
Today: AI Job Hunter analyses the job description. It shows the knockout requirements, the Required and Preferred keywords, and the key duties, each with the line from the job description it came from. Tailoring your CV to those keywords is up to you.
Coming next: coworking the keywords into your CV correctly, together with you. We won’t write a fake CV for you with hallucinations. You stay in control. We’ll also show how your CV scores against the ATS for the job, so you can see your progress as you improve it.
Good to know
- Each ATS works a little differently, and an employer can adjust the scoring for each job posting.
- AI Job Hunter is an AI-enabled service. It can be wrong or miss details. Check the results against the job description.
How AI is used
The job description text is analysed through many targeted requests to AI models from third-party providers, and our own rules and code then check the result against the posting. Full details are in the Privacy notice.
Sources
- Filtering and ranking: Harvard Business School and Accenture, Hidden Workers: Untapped Talent. Over 90% of the employers surveyed use their recruiting system to filter or rank candidates.
- How few are read: Ashby, 2026 Talent Trends Report. Across 109 million applications, 3.6–4.7% of applicants reached an interview.
- LLM comparison: our own test of 20 October 2026, five job postings, one run each.