Scouting

Market · 7 min read · 4 February 2026

How AI is changing hiring in the UAE

Every open role in Dubai draws a crowd. The interesting change isn't that software reads CVs now — it's that the reading has to be explainable.

Post an operations role in Dubai on a Sunday morning and you will have 200 applications before lunch on Monday. Post a marketing role and it's worse. Part of that is the size of the talent pool here: a genuinely global applicant base, a lot of people already on visitor visas hoping to convert, and a habit of applying to everything that moves because applying is free. The volume isn't a sign of a broken market. It's just the market.

What has changed in the last two years is what happens to those 200 CVs. They used to be skimmed by whoever had the least on their plate — an office manager, a junior in HR, sometimes the hiring manager at 11pm. Now they get parsed, scored and ranked before a human opens a single file. The part people argue about is whether that's an improvement. My view: it is, but only when the score comes with a reason attached.

The old filter was worse than we admit

Manual screening in a high-volume market isn't careful. It's a series of shortcuts under time pressure. Nice PDF template, familiar employer name, a school someone recognises, a gap they can't explain in four seconds. Those shortcuts favour candidates who know the conventions of the local market, which usually means candidates who are already inside it. A brilliant logistics supervisor from Karachi with a plain Word CV and a ten-day notice period loses to a mediocre one with a designer template and a LinkedIn banner.

So when someone says automated screening might be biased, the honest reply is: compared to what? The baseline was a tired person at 11pm making pattern-matching decisions they couldn't explain the next morning. The bar for improvement is lower than the debate suggests.

What the software is actually good at

Two things, mostly. First, turning unstructured text into comparable fields — years of relevant experience, tools, languages, notice period, salary expectation, whether someone has worked in this region before. Second, comparing those fields against a job spec consistently, so candidate number 194 gets the same treatment as candidate number 3. Consistency is unglamorous and it's the entire point.

What it is not good at is judging the things you hire for at the end: whether someone will stay two years, whether they will cope with a family-owned business where the founder changes his mind on Thursday, whether they can hold their own with a difficult supplier in Jebel Ali. Nobody should be pretending otherwise. Screening narrows 200 to 10. Humans still pick the one.

Scores without reasons are just a new black box

This is where a lot of tools fall down. A dashboard tells you candidate A is an 87 and candidate B is a 61, and you have no idea why. You can't defend that to a hiring manager, you can't sanity-check it, and you certainly can't explain it to a candidate who asks. Worse, you stop noticing when the scoring is wrong, because there's nothing to disagree with.

A score needs a sentence. "Eight years in third-party logistics and fluent Arabic, but no experience managing a team above five" is something a human can argue with. That argument is the useful part: it's where you learn that your spec asked for team management you don't actually need, or that you do need it and should say so louder.

What this means if you're job hunting here

Write for extraction, not for decoration. The first reader of your CV is going to pull out fields, so make the fields easy to find: job title, employer, dates, what you were responsible for in numbers, tools you actually use, languages, notice period, where you're based and whether you need a visa. A two-column design with icons and a skills bar chart doesn't help you now. Clear text does.

And stop tailoring by adjective. Tailoring means changing the specifics — swapping which projects you lead with, putting the relevant tool in the first line of the role rather than the last. It doesn't mean adding "dynamic" and "results-driven" to the top.

What this means if you're hiring here

Write the spec like you mean it. Vague specs produce vague shortlists, and that's true of software and humans equally. If the salary band is 18 to 22, say so; half the churn in UAE hiring is two parties discovering in week three that they were never in the same range. If Arabic is genuinely required, mark it required. If it's nice to have, don't let it quietly disqualify half your pool.

Then treat the shortlist as a starting point you're allowed to push back on. The teams getting the most out of this are the ones who read the reasons, notice a pattern they disagree with, fix the spec, and re-run. That loop takes ten minutes and does more for quality of hire than another job board subscription.

Scouting scores every candidate against every role you post, 0 to 100, with the reason written out in plain language. You can post a role in about two minutes and see what the pool looks like before you commit to anything.

Try it on a real role

Scouting scores every candidate-role pairing 0 to 100 and writes out the reason in plain language. Companies unlock a candidate's full profile and standardised report when they're ready to talk.

Keep reading