Zero Assist

Product Introduction

1. Definition: Zero Assist is a forensic-grade, real-time AI cheating detection and interview integrity monitoring platform. It operates as a silent desktop agent that monitors a candidate's operating system during remote technical interviews.
2. Core Value Proposition: It exists to provide hiring teams with forensic certainty in remote technical assessments by detecting the unauthorized use of AI-powered cheating tools like Parakeet AI, Final Round AI, and ChatGPT, which bypass traditional video proctoring and screen-sharing methods.

Main Features

1. UIA/AppleScript URL Extraction: This feature bypasses window-title obfuscation to extract real-time browser URLs directly via Windows UI Automation (UIA) and macOS AppleScript. It detects when a candidate navigates to AI tool websites like chat.openai.com, even if the tab or window title is hidden or renamed.
2. OS-Level Process & Overlay Monitoring: The agent performs deep system scans to identify background processes of known AI desktop applications (e.g., Parakeet AI, Claude desktop app) and detects stealth screen overlays by analyzing z-index values and window properties, which are invisible to standard screen-sharing software.
3. Real-Time Behavioral Telemetry & Alerting: A live dashboard provides interviewers with instant WebSocket-pushed alerts (under 500ms latency) when suspicious activity is flagged. This includes process matches, network DNS queries to blacklisted domains, and anomalous clipboard operations, complete with a live "integrity score" gauge.
4. In-Session Whitelisting: Interviewers can authorize legitimate tools (e.g., Stack Overflow, official documentation) mid-interview with one click. This refines the detection scoring without stopping the session, distinguishing between allowed research and prohibited AI assistance.
5. Forensic Audit Trail: Every session generates a comprehensive, timestamped log of all monitored events, including evidence captures (e.g., screenshots of flagged overlays) and process activity. This creates a compliant audit trail for post-interview review and candidate dispute resolution.

Problems Solved

1. Pain Point: The "integrity gap" in remote hiring, where candidates can use hidden AI tools, screen overlays, and automation scripts that are completely undetected by standard video conferencing and screen-sharing software, leading to fraudulent hiring decisions.
2. Target Audience: Technical hiring managers, engineering leads, technical recruiters at tech companies and unicorn startups, and remote assessment platforms who conduct live coding interviews, system design interviews, or take-home challenges with real-time components.
3. Use Cases: Essential for securing remote interviews for software engineering, data science, DevOps, and any technical role where candidates might use AI coding assistants (like GitHub Copilot in an unauthorized context), real-time answer whisper tools (like Parakeet AI), or overlay tools to receive external assistance during a live evaluation.

Unique Advantages

1. Differentiation: Unlike basic proctoring software that only analyzes video feeds or screen content, Zero Assist operates at the OS level, the only layer where modern AI cheating tools cannot hide. It does not rely on invasive constant screen recording but on forensic system telemetry.
2. Key Innovation: Its dual-layered forensic approach combining UIA/AppleScript for application-layer URL detection with OS-level process scanning for system-layer activity. This allows it to catch both browser-based AI queries and background desktop AI applications simultaneously, a method that bypasses obfuscation techniques used by stealth AI tools.

Frequently Asked Questions (FAQ)

1. How does Zero Assist detect AI cheating tools like ChatGPT? Zero Assist uses a lightweight local agent to monitor the candidate's operating system in real time. It cross-references running processes and network activity against a continuously updated database of signatures for over 20 AI cheating tools, including ChatGPT Desktop, Parakeet AI, and Cluely, alerting the interviewer instantly upon a match.
2. Is the Zero Assist agent spyware or malware? No, it is not spyware. The agent is a single-session, forensic monitoring tool that requires candidate consent to run, does not require administrator privileges, captures no persistent personal data, and terminates completely at the end of the interview session. It is designed for specific, transparent interview integrity purposes.
3. Can candidates bypass Zero Assist using a virtual machine (VM) or remote desktop? The platform includes detection signals for virtual machine environments and remote desktop software, which are common flags for sophisticated cheating attempts. These anomalies are reported to the interviewer as high-severity warnings within the live dashboard.
4. What happens if a candidate uses an allowed tool like Stack Overflow? Interviewers can whitelist approved websites and applications (like Stack Overflow or official documentation) during the live session. Once whitelisted, activity to these resources will not trigger alerts, allowing the platform to focus solely on detecting unauthorized AI assistance.
5. Is Zero Assist compliant with data privacy regulations like GDPR? Yes, the platform is built with compliance in mind, featuring configurable data retention policies, automated evidence deletion, EU/IN data residency options, and is undergoing SOC 2 Type II certification. It collects only the forensic telemetry necessary for interview integrity.