Headroom vs InterviewWatch

It also checks your driver for the subgroup bug that makes some in-browser LLMs produce garbage.

The InterviewWatch agent flags AI assistance, hidden overlays, extra displays and remote control during a live interview, then signs a tamper-evident report.

About Headroom

It also checks your driver for the subgroup bug that makes some in-browser LLMs produce garbage. One outbound request: ~0.4 MB of public model metadata to derive bytes per token, no run data attached. Projections are measured read bandwidth ÷ bytes touched per token, at empty context; derived manifests make bytes-per-token exact, estimates fall back to an honest band. Real engines land below the ceiling; the gap is engine overhead (see methodology). Runs the transformers.js WebGPU path on the exact model your ceiling is quoted for ( onnx-community/gemma-4-E2B-it-ONNX.

About InterviewWatch

The InterviewWatch agent flags AI assistance, hidden overlays, extra displays and remote control during a live interview, then signs a tamper-evident report. An agent runs on the candidate's machine during the interview and flags AI assistants, hidden overlays, remote control and virtual devices as they appear. Why it matters Interview cheating stopped being an edge case. If your interviews aren't monitored, you may already be extending offers on answers that were never the candidate's. Across 19,368 technical interviews, 38.5% of candidates were flagged for AI-assisted cheating, and the rate climbed from 9% to 45% in a single quarter.