Everything it takes to make an interview count as evidence
Luminos Hire runs the interview, records what happened in it, and puts it in front of a person who decides. Here is every part of that — including the parts still on the roadmap.
- Capabilities
- 20
- Trust tiers
- 3
- Separate signals
- 3
- Languages
- 5
Questions that follow the answer, not a script
The engine plans a route through the skills a role needs, then rewrites it live: a thin answer earns a follow-up, a strong one moves on. Every turn is scored on its own, so the transcript shows where a conclusion came from.
- Spoken answers, transcribed turn by turn.
- A real coding exercise in the browser, for the roles that need one.
- A dropped connection resumes where it stopped — no restart, no lost session.
Consented, monitored, and written down as it happens
Nothing records before the candidate agrees, and the recording indicator stays visible for the whole session. What the session emits is a timestamped log of observations for a reviewer to read — not a risk score, and not a verdict.
- Identity attested at the start of the session.
- The candidate's phone pairs by QR as a second camera.
- Every observation carries a time, so a reviewer can go straight to it.
The report card a client receives
A reviewer works through a checklist against the transcript, then writes a conclusion in their own words. The badge tier, the skill breakdown and that conclusion arrive as three separate things, with the transcript attached so you can check any of them.
- A reviewer checklist, applied the same way every time.
- The full transcript, attached to the scores it produced.
- Downloadable as a PDF, for the people in the room who are not in the tool.
The full list
Grouped by what it does for you. Anything not shipped yet is marked as such.
Run the interview
A structured session that adapts to the person in it — in any domain, not just technical roles.
Adaptive question engine
Questions are planned from the skills a role needs, then adapted live. A thin answer earns a follow-up, so the number of questions is never fixed in advance.
In-browser coding exercise
A real editor with a runnable exercise for technical roles, scored beside the spoken answers instead of in a separate tool.
Spoken answers, transcribed
Candidates answer out loud. Each turn is transcribed and scored on its own, so every conclusion traces back to something that was actually said.
Interview templates per role
Build a template from a job description, reuse it across candidates, and compare like with like.
Resumable sessions
A dropped connection, a dead battery or a closed tab resumes at the question it stopped on, with the session intact.
Prove the session
Consent first, then evidence: who was there, and what happened, recorded neutrally.
Consent before anything records
The candidate reads what will be captured and agrees before the session starts. The consent is timestamped and stored with the session.
Identity attestation
The person in the session is checked against the invite at the start, so the interview is attached to a candidate rather than to a link someone forwarded.
Phone as a second camera
The candidate pairs their phone by QR for a desk-level angle the laptop camera cannot show, with an on-screen guide for placing it.
Session integrity signals
Neutral, timestamped observations — device pairing, focus changes, reconnects — recorded for a reviewer to read in context. Never a score, never an accusation.
Kiosk-verified sessions
Coming soonAn on-site session at an attested device and location, for the highest trust tier. On our roadmap, not yet available.
A person decides
Every badge, score and recommendation passes through a human reviewer before anyone sees it.
Client report card
One document: badge tier, skill breakdown, strengths, areas to review and the reviewer's conclusion — as separate sections, downloadable as a PDF.
Human review console
Reviewers work the transcript against a fixed checklist, so two candidates are read the same way by two different people.
Full session transcript
Every question, answer and score in one place, each linked to the moment it happened.
Human-granted badges
The verification badge is assigned by a reviewer at one of three tiers. No badge is ever issued by the model alone.
Fit your pipeline
Invites, matching, retention and roles that slot into the hiring process you already run.
Interview by link
Send an invite and the candidate joins in the browser — no account, no install, no scheduling back-and-forth.
Explainable matching
Match talent to a requirement by domain and skill fit, with the reasoning shown and a person curating the shortlist.
Interview analytics
Completion, duration and score distribution across your interviews, so the process itself can be tuned.
Erasure and retention
Candidates can request erasure of their data, and sessions and evidence expire on a retention window rather than sitting forever.
Five languages
The public site and the candidate experience in English, German, French, Dutch and Spanish — detected from the browser, switchable at any time.
Roles and permissions
Client and internal ops teams are separate, each with its own permissions, so a reviewer's access is not an administrator's.
What we deliberately don't do
A feature list is only worth something next to the things the product refuses to do.
- We never publish a machine's verdict on a person. The AI structures the interview and summarises it; a named reviewer writes the conclusion.
- We never merge the three signals into one number. A trust badge is not a skill score, and neither of them is a hiring decision.
- We don't log keystrokes, read the candidate's files, or watch anything outside the session window. Recording is visible for its whole duration.
- We don't claim to detect dishonesty. We record what happened and hand it to a person — a smaller and more honest claim than the one the category usually makes.
If a claim can't be sourced or shown, it doesn't go on this page.
Questions people ask before starting
What does it take to run the first interview?
An invite. You describe the role, Luminos Hire builds the interview from it, and the candidate joins from a link in their browser — nothing to install on either side.
What does the candidate have to do?
Agree to what is recorded, join in a browser, and — at the supervised tier — point a phone at the desk. There is no account to create and nothing to download.
Does this replace our applicant tracking system?
No. Luminos Hire runs the interview and produces the report; it is built to sit inside the hiring pipeline you already have, not to become it.
Which languages are supported?
English, German, French, Dutch and Spanish, detected from the browser and switchable at any time. Adding a language is a translation file, not a rebuild.
Separate product
Recorded interview evidence with Proctor
Proctor is a separate interview-integrity product. A consent-based desktop agent collects only the session evidence a candidate agrees to, and your reviewers decide what it means. It suits teams that run interviews in their own tools rather than on Luminos Hire.
- Separate accounts: signing in to Luminos Hire does not sign you in to Proctor.
- Signals assist and people decide — no automatic outcomes.
See it on a role you're actually hiring for
Send us the role and we'll run a verified interview against it, end to end.