Finding candidates

AI Recommendations

Jobdrive searches its candidate database for resumes that match your Job Description and ranks them for you. It finds people who are a good fit for the role but have never applied to it.

What it is

Every other way of getting candidates into a Job Drive starts with the candidate. Someone applies, or you upload their resume, or you import a list you already had. AI Recommendations works the other way round: it starts with your role and goes looking.

Jobdrive holds resumes from across the platform. When you ask for recommendations, it compares your Job Description against that pool and hands back a ranked list of the closest matches, each with a score and a written reason. Nobody on that list applied to your role.

Use this when a role has gone quiet. The applications have dried up, the shortlist is thin, and you would otherwise be back on the job boards writing another post.

Where to find it

Open the Job Drive you want candidates for, from the Dashboard or the sidebar.
Go to the Resumes workspace for that drive.
Select the AI Recommendations tab.

Candidates Sheet Upload AI Recommendations

Recommendations are generated per Job Drive, against that drive's Job Description. Two roles will not return the same list.

The sourcing tabs inside a Job Drive.

How the matching works

Matching runs in two passes. The first is fast and casts wide, the second is slow and reads carefully. Together they mean you get a short, considered list rather than a thousand keyword hits.

Pass one

Narrow the field by meaning

Jobdrive compares the meaning of every resume in the database against the meaning of your Job Description, and keeps the closest. This is not keyword search. A resume that says "process validation" can match a role asking for "GMP qualification" without sharing a word.

Pass two

Read the shortlist properly

The candidates that survive pass one are read in full against your Job Description. Each comes back with an AI Match Score out of 100, an AI Rating band, and an AI Reason explaining the score in plain English.

Pass one is a similarity search over the whole database. Pass two is a careful read of only what pass one kept.

Generating recommendations

In the AI Recommendations tab, click Generate Recommendations.
The matching runs in the background and takes 30 to 90 seconds, depending on how much is in the database.
Results appear ranked, highest match score first.

You can run it again whenever you like. A fresh run includes every resume added since the last one, so a role you generated for last month is worth regenerating today.

Reading a recommendation

Each result carries the candidate name, email and phone number, a summary of their resume, the AI Match Score, the AI Rating badge, and the AI Reason behind it.

Ranked recommendations

Priya Nair

7 years, Quality Assurance, formulations plant

Excellent 94

AI Reason Seven years in QA at a formulations plant, with audit experience and the documentation practice this role asks for. Currently one level below the seniority in the Job Description, so the step up is real but small.

Rahul Menon

5 years, Production, bulk drug manufacturing

Good 78

Two recommendations, ranked by match score, each with the reason behind the number.

What the rating bands mean

Rating	Score	What to do with it
Excellent	90 to 100	Very strong match. Start here.
Good	70 to 89	Solid candidate, worth reviewing.
Average	50 to 69	Partial match. Read the reason before deciding.
Poor	Below 50	Low match. Usually not suitable.

The AI Reason is the part worth reading. It states where the candidate is strong and where they fall short of your Job Description, which is often enough to make the call without opening the resume at all.

What you can do next

Action	What happens
Shortlist	Moves the candidate into the Candidates Tab for full review.
Reject	Logs a rejection with an optional reason, kept for your own records. The candidate is not told.
Request Resume	Emails the candidate a secure link asking for an up to date CV. Useful when the resume on file is a few years old.
Invite to Interview	Sends an interview invitation by email or WhatsApp, without going through the Candidates Tab first.

Getting better results

Wait until the database has 10 to 20 resumes before your first run. Matching compares against the whole pool, so a bigger pool finds better people.
Put the real requirements in the Job Description. A vague description produces vague matches, because the description is the only thing the scoring has to work from.
Regenerate after a busy week. Anyone who submitted a resume since your last run is included in the next one.
Pair it with a spreadsheet import. Bring in the contacts you already know, and let AI Recommendations find the ones you do not.

Run it on your own role.

Start free Book a demo