01
Deep proctored screening
Camera, microphone and full-screen capture the candidate agrees to first, a server-owned clock, and auto-submit at the deadline.
Best for · High-volume first rounds
Consent-firstServer clockEvidence trail
Stop hiring candidates who can't actually code.
Identify engineers who can build and maintain real products.
10 standard invitations free · No credit card
No tiers and no sales call.
One browser workspace, three ways to run it — pick per role.
Camera, microphone and full-screen capture the candidate agrees to first, a server-owned clock, and auto-submit at the deadline.
An in-workspace AI assistant with a set budget. The whole conversation is captured, and how well they use it is scored.
A shared editor with synced cursors, author-tagged so interviewer keystrokes never pollute the record. Nothing is auto-scored: your notes and rating are saved to the candidate.
Set your volume and costs. Every figure on the right is a sum you can check.
No sandbox setup. No IT ticket. A recruiter can send their first proctored assessment in the time it takes to make a coffee.
Every screening captures the full work session: keystroke-level code timeline, paste events with internal-vs-external classification, tab focus, environment fingerprint, and continuous camera, microphone and full-screen recording the candidate turns on before the clock starts. The scoring engine turns that into an explainable integrity score where every deduction is named, plus a flags timeline you can scrub against the code replay. It is evidence, never just a number.
The dual score. DeepAssessment grades the code AND how it was produced — a Core Coding score from hidden weighted test cases plus an AI review, and an AI Augmentation score for candidates you allow to build with AI. Incumbent platforms tell you tests passed; they cannot tell you whether the candidate or ChatGPT passed them.
Flip the AI toggle on any assessment. The candidate gets an AI assistant inside the test workspace with a set budget, the entire conversation is captured, and the report scores their collaboration — prompt quality, whether they verified AI output, how much they refined it. You see how they code and how they build with AI, side by side.
Industry estimates put one wrong senior hire at $250,000 or more — salary, recruiting fees, lost momentum, and starting the search over. A DeepAssessment screening costs $5. The evidence report exists so you get the signal before the offer, not six months after.
A clean browser IDE — no installs, no account required, one link. Everything captured is disclosed on a consent screen before the clock starts, a recording indicator stays visible for the whole session, and their work is auto-saved so a dropped connection never loses progress.