Contact usRequest a Demo
AI tool detection

Catch AI Assistance That Screen Share Can Miss.

InterviewWatch detects AI tools and AI-assisted interview workflows at the system level: visible browser tabs, dedicated desktop apps, local LLM runners, coding assistants, AI DNS traces, and stealth helpers designed for live interviews.

CoverageDetection methodsFAQ
ChatGPT tabActive
Cursor / WindsurfReview
Local LLM runnerPresent
AI DNS traceWeak signal
What we detect

From mainstream chatbots to interview-specific assistants.

The goal is not to chase one brand name. InterviewWatch combines built-in signatures with remotely updateable tool catalogs so new AI assistants can be covered without waiting for a full agent redeploy.

AI chat and search

ChatGPT, Claude, Gemini, Copilot, Perplexity, AI search pages, and browser-based assistant activity.

Coding assistants

Cursor, Windsurf, GitHub Copilot-style workflows, local LLMs, LM Studio, Ollama, and similar developer tools.

Interview helpers

Purpose-built interview assistants, hidden overlays, answer-injection tools, and stealth windows.

Detection methods

Several surfaces, one reviewable finding.

AI use is strongest when multiple signals align: a known AI process, a matching window title, an AI-assistant DNS lookup, a focus shift, and a suspicious paste or answer burst.

Process and title matchKnown AI app or browser tab appears during the session.
Strong
Network/DNS traceBrowser assistant domain appears after baseline.
Support
Input patternLong pause followed by a fluent structured answer.
Context

Does this read what the candidate types?

No. The detection uses metadata such as tool presence, window context, event timing, and clipboard size/source. It does not capture keystroke content.

What if AI tools are allowed?

Teams can treat the finding as context instead of a failure. The value is visibility: the report shows whether AI was used during a round where the policy matters.