home / use cases / documents and vision / cv-screening

CV screening that keeps candidates private

An Aivell reads every application that lands in your jobs mailbox, scores it against the role's must-haves and nice-to-haves, keeps a ranked shortlist with a short written reason per candidate, and acknowledges every applicant politely. Candidate data is processed on a box in your office — nothing is sent to any cloud service.

Watch the demo

starts when: A candidate emails a CV to your jobs@ mailbox

The problem

Post a decent job ad and a hundred applications arrive in a week. Screening them is repetitive reading under deadline: the same job description held against a hundred differently formatted lives, most of them clearly not a fit, a few of them excellent and buried at position seventy. By the time the pile is read, the best candidate has accepted somewhere faster.

The obvious fix — cloud screening tools — has an ugly cost. Applicants trusted you with their life story, their address, sometimes their salary history. Piping that to a third-party platform is the kind of processing that keeps data-protection officers up at night; reading it on your own hardware is privacy by design, in the literal GDPR Article 25 sense.

How your Aivell does it

  1. The application arrives, the reading starts. No forwarding, no export — the trigger is the email itself, at whatever hour the candidate hit send.
  2. The CV is read on the box. PDF, Word file, scan or photographed paper résumé — a vision model onboard reads them all, and no candidate document ever leaves your office.
  3. Scored against your criteria, not generic ones. Must-haves are checked first — the licence, the language, the permit — and a missing one is named, not silently penalized. Nice-to-haves shape the score above that line.
  4. The shortlist writes itself. Every candidate lands in the sheet with a score and a two-line reason a human can audit. Standouts are flagged to your recruiter the same day, because speed wins hires.
  5. Candidates hear back, humans decide. Each applicant gets a courteous acknowledgement in their own language — and that is the only mail your Aivell can send them. Interviews, rejections and offers stay with people.

The scoring rules are a paragraph of plain language you can read, question and change. That, plus a reason logged for every score, keeps the process explainable to a candidate, an auditor — or yourself, three months later.

The chore, as you'd write it.

No flowcharts, no code — a chore is just a message to your Aivell, in your own words. It shapes it into a solid, guarded procedure and follows it to the letter. This one:

Giulia Aivell · screen incoming applications ● active

watch jobs@ for new applications

read the cv and cover letter, whatever the format

score against the role in the subject line — must-haves first, nice-to-haves second

a must-have missing — licence, language, work permit? mark not-a-fit and say which one

add every candidate to the shortlist sheet with score and a two-line reason

scores above 8 — tell Nadia the same day, not in the weekly batch

send each applicant a polite acknowledgement in their own language

never write a rejection — the sheet ranks, people decide

You
Got it — I'll start right away, and check with you whenever something needs a human.
Giulia Aivell

The tools it uses.

Connected with single-click passkeys — no copied tokens, no OAuth hell.

Email collects applications, sends acknowledgements
Documents & Vision reads CVs in any format, onboard
Spreadsheets keeps the ranked shortlist current

Sensible guardrails.

Every tool can be limited to exactly what this chore needs. For this one, you might set:

Email May send acknowledgements to applicants. Rejections and offers are not messages it can write.
Spreadsheets Writes to the shortlist sheet for the open roles; the rest of the HR share is off-limits.

Questions, answered.

Where does applicant data go? +

Into your own shortlist sheet, and nowhere else. The reading and scoring happen on the box inside your network — no CV is uploaded to a screening service or an AI provider. When a hiring round ends and you delete the data, it is actually gone, because it only ever existed on your systems.

Does the AI reject candidates? +

It cannot. A guardrail limits its outgoing mail to acknowledgements, so a rejection is a message it is physically unable to send. It ranks, explains its ranking in two lines per candidate, and leaves the decision to the humans reading the sheet.

Can we check why a candidate was scored low? +

Always — the reason is written next to the score, in plain words, for every applicant. If you disagree with a pattern ("stop discounting career gaps"), you change the chore and the next applications are scored your way.

What file formats does it handle? +

The usual PDFs and Word files, but also exported profile pages, scans, and the occasional photographed paper CV — the vision model on the box reads those too.

Related.

What is an Aivell?

An Aivell is an on-premise AI colleague: a small box that plugs into your network, runs its own AI onboard and takes the repetitive work off your desk. Prompts, documents, data — nothing ever leaves your office. Unplug the internet: it keeps working.

You describe each task as a chore, in plain language. Your Aivell turns it into a solid, guarded procedure and runs it in the background — with guardrails on every tool, a complete audit trail, and approvals in your hand when you want them. One-time setup, fixed monthly fee. No tokens, no overages.

Watch the demo

onboard inference · air-gapped · fixed monthly fee