Wafaqli

Hiring agent that tells you why you were rejected

September 6, 2026 I'm a doctor. I built a hiring agent that tells you why you were rejected.

I'm a cardiac anaesthesia consultant. I've built free clinical decision-support tools before — one is published in a peer-reviewed journal, another reaches 157 countries. This is my first product outside medicine, and I'd like it broken rather than praised.

The problem: someone looking for work opens five sites, retypes their CV into each, and sends dozens of applications without ever learning why any was refused. Someone hiring gets a hundred applications they can't read, so they read the first ten. Both sides lose for the same reason — volume replaced reason.

So Wafaqli does the opposite. Fewer results, each carrying a written reason: what fit, what was missing, what is unknown. A mandatory requirement stays separate from a preferred one, because a missing must-have is a no, not a deduction from a score.

Search runs in three layers and the order is binding: vacancies inside the platform first, then a live search of the open web when there aren't enough, then possible employers — which are never called vacancies. External results keep their source and the publication date the page itself states, or "not stated" when it states none.

Some constraints are written into the code and held by automated tests rather than by promises: no bulk apply, nothing sent in anyone's name without approval, no protected characteristics used in matching, no automatic rejection, and "unknown" never passes as a pass or a failure.

Where it actually stands: launched days ago, few internal vacancies, no company has posted a real job yet, and the live prompt-injection test isn't done — the API key wasn't scoped to a workspace and I'm not going to claim I ran it.

Two things I want to be argued with about:

1. Is "a reason behind every match" a real differentiator, or does it just sound like one? Every matching tool claims explainability now.

2. The job seeker side works today because layer two returns real results. The employer side is an empty database. Is a two-sided product with one live side worth launching, or should I have waited?

Arabic-first, four other languages. Free during launch.

Ahmed75

2 Likes

2 Comments

Say something nice…

Post Comment

1

The explanation sounds differentiated, but the stronger test is behavioral: do job seekers actually change which roles they pursue because of the reason, rather than simply appreciating the explanation?

Aryan Sinh

·
4 hours ago
·
Reply
1. 1
  
  That's the sharper version of my own question, thank you. You're right that appreciation is cheap. The behavioural test I can actually instrument is narrow: does someone who sees a gap go and close it, or drop that role and pursue a different one? Both are behaviour change. Both are measurable as events. Neither is measurable yet — the analytics only track the funnel to application, not what happens to the gap afterwards. One thing I'd push back on slightly: there's a second behaviour that doesn't need the seeker to change anything. The reason exists on the employer's side too, next to each candidate. If it changes who gets read past the first ten, that's behaviour change on the side that was never going to appreciate an explanation. Both are unproven. I'll take the criticism that I launched claiming a differentiator I can't yet demonstrate.
  
  Ahmed75
  
  ·
  4 hours ago
  ·
  Reply

Wafaqli

About

People