AI pre-screening: every applicant gets an answer

Candidates hate applying into a void. Recruiters hate the void too. They're just buried under hundreds of applications per role, with no time to reply to everyone.

I built two small tools to fix both sides.

Tool one: the pre-screening chat

Before applying, candidates have a short conversation with an AI recruiter that has read the job description and their CV. It isn't trying to catch anyone out. It looks for deal-breakers: a missing core skill, a location that won't work, contract terms that don't fit.

I tested it on myself. Applying for a back-end Java role with a JavaScript CV, the interviewer asked fair, specific questions about Java, AWS, PostgreSQL and Kotlin, then declined kindly and with clear reasons. Applying for a front-end role, it picked up a subtle mismatch (I wanted remote, they said hybrid), asked about it, accepted "occasional visits to Gloucester, maybe once a month," and moved the application forward.

Both candidates walked away knowing exactly where they stood.

Tool two: the bulk CV processor

Paste in a job description of any length, choose what to screen on (location, skills, experience), drop in the CVs and press go. Each takes a few seconds, so 500 applications take roughly three to five minutes.

Every candidate gets:

• A one-to-five rating
• A verdict: good match, weak match or not matching
• A short, specific explanation

In my test of 44 CVs, 21 matched and 23 didn't. The reasons read like a sharp recruiter's notes. One rejected candidate was a Python machine learning engineer applying for a JavaScript role. Another lived abroad with none of the listed skills. One strong match lost a star for not mentioning a skill the role asked for, and the summary said so.

Doing this responsibly

Hiring decisions affect people's lives, so the rollout matters as much as the model.

• Start with the obvious. Location, right to work, contract type and availability are low-risk, objective filters, a sensible first step before anything involving judgement.
• Explain everything. Every decision comes with a reason a human can check and a candidate can understand.
• Keep humans in the loop. The tool sorts and summarises. People decide.
• Test for bias before and after go-live, on your own historical data.

Who this helps

Recruitment agencies, in-house talent teams and job boards. Anyone drowning in applications who'd like to give every candidate a fast, fair, specific response.

I build AI screening around your process, your ATS and your compliance needs. Let's talk about your hiring pipeline.

More on this project

• The demo on YouTube
• An AI assistant that takes recruiter calls