Selekta
boltBuilt for Indonesian HR teams

Stop reading CVs.
Start reading shortlists.

Drop a batch of CVs in. Selekta extracts the data, scores every candidate against the role on three dimensions, and hands you a ranked shortlist with the reasoning attached. Your data never leaves Indonesian infrastructure.

verified_user UU PDP aligned · hosted in Jakarta · 12-month retention by default
SW
Sari Wijaya
Senior Backend Engineer
starStrong Match
Related Skills91
Experience Level88
Education84
Overall
88
AI verdict
Deep backend and cloud experience that maps closely to the role, with a matching degree.
15
CVs screened in parallel
~40s
Per CV, extract to score
3
Scored dimensions per role
100%
Actions written to an audit log
The workflow

Four steps, one upload

You do step one. Selekta does the rest and tells you when it is finished.

STEP 01
upload_file

Upload a batch

Drag in every CV for the role at once. PDF only, so nothing weird slips into the pipeline.

STEP 02
document_scanner

Extract

OCR pulls the text off the page, then structured fields come out: contact, experience, education, skills.

STEP 03
grading

Evaluate

Each candidate is scored against your job description on three dimensions, with a written verdict.

STEP 04
checklist

Shortlist

Sort, filter, move candidates through the pipeline. Rejections and invites go out on your rules.

Try it

How a score becomes a decision

Every candidate lands in a band. The band is what your team actually acts on, and it is calculated the same way every single time. Move the slider.

0100
Overall score
88
Strong Match
Pipeline

One place the whole hire lives

Stage changes are logged with who did it and when. Nothing moves silently.

New arrow_forward Under review arrow_forward Shortlisted arrow_forward Hired alt_route Rejected

See it run on your own CVs

Bring a real job description and a stack of real applications. Thirty minutes is enough to know whether this fits your team.

calendar_monthBook a walkthrough
Features

What you get

Everything below is in the product today, not on a roadmap slide.

stacks

Batch screening

Upload the whole applicant pile in one go. Fifteen CVs process at a time and a bad file never stalls the rest of the batch.

query_stats

Multi-dimensional scoring

Related Skills, Experience Level and Education, each 0 to 100, equally weighted into one overall score. No hidden weighting to argue about.

psychology

Written verdict

Every score comes with a one-sentence justification and a summary, so a hiring manager can see the reasoning instead of a bare number.

person_search

Candidate database

One person, many applications. Search across every role you have ever posted and see a candidate's full history in one view.

mark_email_read

Rule-based candidate email

Set a rule per stage, with a delay in business days. Weekends are skipped and mail only goes out inside working hours.

filter_alt_off

Auto-reject threshold

Set a score floor per job. Anything under it is rejected on arrival, logged separately from manual decisions, and never touches a candidate you already moved forward.

quiz

Interview questions

Generate questions tailored to the gaps in a specific candidate's CV, on demand, and reuse them at the interview.

download

Excel and PDF export

Take the shortlist to a meeting in a format everyone already opens. Every export is recorded in the audit log.

history

Audit trail

Stage changes, exports, deletions, logins. Who did what, when, from where, kept for when someone asks.

Ready to see it on real CVs?

We will run your own applications through the pipeline and walk you through the output.

rocket_launchRequest a demo
Data & Privacy

Where candidate data lives

Screening CVs means handling personal data. Here is the honest version.

Residency

Application records, CV files and extracted candidate data are stored on cloud infrastructure in Jakarta, Indonesia. Text recognition runs in Singapore because that capability is not offered in the Jakarta region.

Consent

Candidates applying through a public careers page pass an explicit consent gate before a CV is accepted. Consent is recorded with a timestamp on the application.

Retention

Applications and their files are deleted automatically twelve months after the last activity on them. Deletion removes the stored file, not just the database row.

What a candidate can ask for

  • A copy of the personal data held about them
  • Correction of anything recorded incorrectly
  • Erasure of their application and CV file
  • Withdrawal of consent, which triggers erasure

What the AI does and does not do

An AI model reads the CV text and produces scores plus a written justification against the job description. It does not make the hiring decision. A person moves every candidate between stages, except where your own team has configured an automatic rejection threshold, and that action is logged distinctly so it is always visible as automated.

Rejection notices

A rejection email never discloses a score, a ranking or an AI-generated reason. Candidates are told the outcome, not the internal assessment.

Access control

Every record is scoped to one company. Administrator-only actions, including screening rules and thresholds, are enforced on the server, not hidden in the interface. Database connections are encrypted and certificate-verified.

Contact

Let's talk

Tell us the roles you hire for and roughly how many CVs land per opening.

mail
Email
hello@noveel.net
schedule
Hours
Monday to Friday, 09:00 to 17:00 WIB
location_on
Region
Jakarta, Indonesia
Common questions
No. Selekta is the pipeline. You can export to Excel or PDF at any stage if another system needs the data.
PDF only, deliberately. It keeps extraction predictable and stops unexpected formats reaching the processing pipeline.
The three dimensions are fixed. What differentiates one role from another is the job description, which is fed into the evaluation directly. Edit the description and the criteria are regenerated to match.
No. CVs are sent to a hosted model for scoring and are not retained or used for training by us or the model provider.