Platform · Screening & scoring
A scored pipeline you can actually open up.
Orbis screens every applicant through four visible stages, knockout filters, semantic relevance match, AI analysis, and an AI video interview, each with a score you can inspect. You set the weights and filters; no hidden model decides who gets through.
The four stages, in detail.
Knockout filters
Rule-based checks on must-have skills, experience, education and location. Imported or low-confidence Arabic resumes are protected by a fail-safe that won’t auto-reject on unreliable data.
Relevance match
Semantic-embedding scoring across five weighted dimensions (skills, experience, education, projects and seniority fit), not just keyword matching.
AI analysis
A deeper qualitative read of trajectory, growth and achievement quality, with red-flag detection, surfaced as inspectable grades.
AI video interview
A recorded AI video interview, scored by a panel of three independent AI models that majority-vote to reduce single-model bias.
You stay in control.
Set the scoring weights across the five relevance dimensions, define your filtering tolerances, add your own interview questions, and review the score behind every decision. Orbis does the heavy lifting; you make the call.
FAQ
How screening works
How does Orbis screen and score candidates?
- Every applicant moves through four visible stages: rule-based knockout filters, a semantic-embedding relevance match across five weighted dimensions, a deeper AI analysis of trajectory and achievement, and an AI video interview. Each stage produces a score you can open and inspect.
Is it a black box?
- No. You set the scoring weights and filters, and you can inspect the score behind every stage for every candidate. Orbis is designed for transparency, not an opaque verdict.
How does it handle Arabic resumes?
- Orbis natively parses and preserves Arabic and bilingual resumes, and a fail-safe avoids auto-rejecting candidates whose resumes parsed with low confidence, so strong people aren’t lost to a parsing error.
Does screening use protected attributes?
- No. Scoring is based on role-relevant signals. A gender field, if collected on the apply form, only flags a mismatch against a role preference, it never disqualifies, hides, or reorders a candidate.
See your pipeline, scored end to end.
Book a 20-minute demo and watch the pipeline screen, interview, and score a candidate end to end.
Book a demo