How Skillytal's interviews, evaluation engine, and proctoring actually work — including what they don't claim to do. Can't find what you need? Talk to our team.
Product & Platform
Skillytal is an AI-powered interview, assessment, and proctoring platform for hiring teams. It runs AI live interviews and structured assessments, evaluates candidate responses across multiple dimensions, and — where the architecture supports it — checks whether the assessment session stayed intact, turning all of it into a single, reviewable candidate report instead of hours of raw transcripts.
Four formats, usable on their own or combined into a single assessment: AI live interviews for natural, conversational conversations; multiple choice for fast knowledge checks; technical questions for domain-specific depth; and scenario-based questions that present a realistic situation — like a production incident or a conflicting stakeholder request — to surface reasoning and prioritization under pressure.
The AI conducts the interview as a conversation rather than a fixed script. Each follow-up question is chosen dynamically based on what the candidate just said, the requirements of the role, and the skill level demonstrated so far — so a strong answer gets probed deeper, and a thin one gets a clarifying follow-up, rather than every candidate hearing the exact same list of questions in the exact same order.
Yes. Assessments are configured per role — which formats to run, which technical topics or scenarios to include, and which proctoring checks apply — so a senior backend role and an entry-level support role don't get the same assessment by default.
AI Evaluation
The evaluation engine reads a response across several dimensions at once — not just whether it's "correct" — and combines those scores into a single, explainable evaluation for that answer. Spoken responses are also analyzed for delivery signals like clarity and pacing, alongside the substance of what was said.
Technical knowledge, problem solving, communication, reasoning, confidence, and response relevance are scored per answer. Across the full session, Skillytal also tracks behavioral signals and assessment integrity, so a candidate's report reflects both what they said and how the session held together.
AI generates the scoring and the candidate intelligence report, but it's built to be reviewed, not to auto-decide. Recruiters and hiring managers see the full breakdown behind every score and make the actual hiring call themselves — Skillytal produces the evidence, not the verdict.
Recruiters get a per-question breakdown in the candidate intelligence report, showing the individual dimension scores and an AI summary behind the overall result — not just a single number. It's built as evidence for a hiring decision, not a black-box score.
Proctoring & Integrity
Where the product architecture supports it, Skillytal runs AI-assisted proctoring directly on the candidate's own device, computing signals on-device rather than shipping raw video off for processing. This keeps the candidate's raw camera and microphone feed closer to them, and surfaces only the resulting signals for review.
It looks for AI-powered signals across vision and audio: multiple people in frame, phones or secondary devices, books or notes, whether the candidate's face stays visible, unusual camera behavior, mismatches between visible mouth movement and detected speech, and audio anomalies. Every one of these is a signal for human review, not an automatic accusation — a flagged item means "look at this," not "this candidate cheated."
No. Skillytal does not claim to catch every possible integrity risk or every AI overlay. Instead, it combines multiple independent signals — camera observations, audio characteristics, speech timing, interaction patterns — into a single, weighted risk indicator, where no single input determines the outcome, and every flagged result is meant for a human to review. An absolute guarantee here would be an irresponsible claim, and we won't make one.
No. Skillytal's audio intelligence measures communication and behavioral signals — words per minute, average pause length, response duration, speech consistency, vocal dynamics, audio anomalies — not emotion or truthfulness. It never claims to determine deception, nervousness, or how someone feels from voice alone, and none of these signals is used as a stand-in for lie detection.
Getting Started
Most teams can configure a role, choose an assessment format, and send an invite link within a single working session. Exact setup time depends on how much customization a role needs — book a demo and we'll walk through it live for your use case.
Book a demo and our team will get you set up to try Skillytal on a real role. That's the fastest way to see how the AI interview, evaluation engine, and proctoring signals behave with your own assessment content.
Pricing depends on team size, hiring volume, and which assessment formats and proctoring features you need. Book a demo or contact our team and we'll put together a plan for your hiring needs.
Recruiters, hiring managers, and technical hiring teams who are evaluating more candidates than they can manually review in depth. Skillytal is built to standardize how every candidate is assessed and to turn hours of transcript review into a structured report a hiring team can actually act on.
Security & Data
Candidate and assessment data is scoped to the organizations and roles that need it, and each assessment link is scoped and time-bound to reduce the surface area for misuse. See the full security overview for how the architecture is put together — we don't claim certifications we haven't earned.
Access is limited to the recruiters and hiring managers on that role within your organization. Proctoring signals and evaluation reports are not shared across unrelated organizations or roles.
Yes. Proctoring is configurable per assessment — enable only the checks a given role actually needs rather than running every signal on every candidate by default.
Talk to our team about your hiring workflow, or see the platform in action.