Skip to main content

What is Sherlock?

In a remote-first hiring world, traditional video interviews are increasingly vulnerable to sophisticated compromise—including real-time deepfakes, unauthorized screen content, and identity fraud.

Written by Gemma from Sherlock

Sherlock acts as an automated integrity engine, providing recruiting and operations teams with the data-driven insights they need to trust their candidates. By analyzing video, audio, and behavioral feeds, Sherlock secures the interview process without adding unnecessary stress or friction for candidates.

During a scheduled session, Sherlock automatically joins the meeting as a participant to observe and protect the integrity of the discussion.


How Sherlock Safeguards Interviews

Depending on your organization's specific security needs, administrators can choose how Sherlock interacts with the candidate and the interview environment.

  • Detect Only (Recommended): Sherlock joins as a participant and quietly observes the session to watch for and report cheating indicators. Candidates do not need to install any software.

  • Prevent + Detect: Active security measures are enabled to block cheating applications and report suspicious background activity. For this mode, candidates will need to install the Sherlock lightweight application prior to the interview. Sherlock will not join as a participant.

  • Realtime Deepfake Detection + Detect: Enables advanced real-time detection specifically for synthetic video, deepfakes, or digital overlays, followed by a comprehensive post-interview integrity report.

Candidate Identity Verification

To prevent identity fraud, administrators can enable Require Identity Verification for Candidate. If checked, Sherlock automatically emails the candidate before the interview, guiding them through a quick step to verify that the person attending the interview is exactly who they claim to be.


1. Detect / Real‑Time Mode Capabilities

  • Behavioral & Environment Monitoring

    • Visual Signals: Identifies visible gaze direction, off‑screen attention, screen‑share context, answer timing, content appearance patterns, and audio cues where observable.

    • Audio Signals: Flags background voices, whispered answers, or unrecognized secondary speakers during the session.

    • Environmental Context: Detects changes in lighting, movement, or background consistency that may indicate external assistance or setup manipulation.

  • Deepfake & Synthetic Video Detection

    • Synthetic Stream Flags: Analyzes live feeds to identify face‑swapping software, digital overlays, or synthetic cameras.

    • Automatic Candidate Identification: Dynamically tracks the candidate’s presence throughout the stream without requiring manual team monitoring.

2. Desktop Mode Capabilities

When Sherlock runs in desktop mode, it focuses on system‑level integrity monitoring rather than video or audio analysis.

  • Application Activity Tracking: Detects unauthorized background applications or browser tabs that may assist the candidate during the interview.

  • Process Monitoring: Flags suspicious system behavior such as screen‑sharing tools, virtual camera software, or external communication apps.

  • Local Environment Verification: Ensures the candidate’s device remains secure and free from tampering throughout the session.


Scoring & Review Workflow

At the conclusion of an interview, Sherlock compiles all detected anomalies into a structured integrity report. The report provides a clear breakdown of trust signals, behavioral flags, and overall risk classification for each candidate session.

  • Integrity Score (0–10): Each interview is assigned a score based on the presence and severity of detected anomalies.

  • Report Insights: The report includes detailed visual indicators of risk level, confidence percentage, and contributing factors such as gaze patterns, off‑screen attention, or background voices.

  • Review Workflow: Interviewers can view, share, or delete reports directly from the sidebar, ensuring quick access to flagged sessions and team‑wide visibility.

Did this answer your question?