Industrial AI practice  ·  Gulf  ·  Arabic and English Reply within one working day
Recruitment document intelligence

Screening applications against a real vacancy

professional-services firm · Gulf

← All documented proof

The shape of this problem

A human reads it and types it in again

Built and run by this practice in its own environment.

The problem

Recruiters needed to compare CVs against job descriptions beyond keyword overlap. Keyword matching rewards the candidates who wrote their CV to mirror the posting rather than the ones who can do the job, and it penalizes anyone who describes the same experience in different words. At volume, the screening pass is also where fatigue does most of its damage, because it is the stage that receives the least time per document and the least scrutiny afterwards.

What we built

We built a workflow that extracted CV content, generated semantic representations, evaluated relevance against the stated criteria, and ranked candidates for recruiter review.

What moved

Measure Before After
Time to a ranked shortlist from a 300-CV posting ~9 hours 25 minutes
Qualified candidates surfaced that keyword screening had ranked below the cut +23%
Recruiter agreement with the top-20 ranking no prior measure 86%

† Measured against the prior level of the same measure.

Time to a ranked shortlist from a 300-CV posting

Before ~9 hours After 25 minutes

Qualified candidates surfaced that keyword screening had ranked below the cut

Prior level 0 100 200 +23%

Measured against the client's own prior process, indexed to 100. The source publishes the size of the movement, not the absolute figure it moved from.

Recruiter agreement with the top-20 ranking

86%

86%

Recruiter agreement with the top-20 ranking

The ring is the whole of that measure. The grey arc is the other 14%.

Figures are from the practice's own delivery records for the engagement named, measured against the process that preceded it.

What it turned on

Ranking for review rather than filtering is the distinction that matters - the recruiter still sees the pool and the reasoning, so the system reorders attention instead of quietly removing people from consideration. Relevant to any high-volume hiring process whose first pass is currently keyword search.

Start here

Which of the four is yours?

Tell us the documents and the monthly volume and we will send the two closest records, with the figures behind each and what it would take to repeat them on your material.

You get a reply within one working day, from the engineer who would do the work.