Question 1

Won't this falsely accuse people?

Accepted Answer

Every integrity flag enters human review with its probability, timestamp and code state attached. Evidence states remain explicit, interview probes stay as questions, and the published thresholds let your team inspect the method directly.

Question 2

What about candidates who use scratch paper, or look away to think?

Accepted Answer

They declare a scratchpad before starting, and looking at it is recorded as note-taking. That is the whole point of the setup step: we compare each candidate to their own baseline — their camera, their lighting, their screen position — instead of a fixed threshold that quietly penalises anyone with a wide-angle webcam or a dim room.

Question 3

How is this different from a coding test with proctoring bolted on?

Accepted Answer

A coding test grades the answer. We keep the process. Every save is compared to the last, character by character, so a session becomes something you can scrub through: the code at any second, what changed, and what the webcam saw at the same moment. Forty minutes with nine iterations and heavy deletion is a different candidate from four minutes with one paste — and if all you stored was the submission, only one of those facts exists.

Question 4

How can we evaluate QualifyCode before switching?

Accepted Answer

Inspect the published thresholds, open the sample session report, and run the same role alongside your current screen. Per-candidate billing lets you compare the evidence each system returns before changing your workflow.

Question 5

Is this legal to use in the EU?

Accepted Answer

QualifyCode records what was decided, on which evidence, under which policy version and by whom. It also provides a four-fifths impact export, explanation pack and identity-preserving erasure. A publishable Local Law 144 audit additionally requires candidate-provided demographic inputs.

Question 6

How does candidate intake work?

Accepted Answer

Type or paste up to 500 candidate emails, review the validation summary and send. Each assessment also carries a reference field so records stay easy to match in your existing workflow.

Question 7

How does the report handle evidence states?

Accepted Answer

The evidence meter identifies each completed, pending or unavailable check. Recommendations require the assessment and code-analysis evidence, while unmeasured dimensions stay outside the score rather than being converted to zero.

Question 8

Which languages and problems are supported?

Accepted Answer

Python is available today with a filterable catalog of dynamic problems and deterministic tests. JavaScript, Java, Golang, Rust and C++ are coming soon, with the same evidence model planned for each runtime.

Question 9

How does billing work?

Accepted Answer

You buy credits and spend one per candidate. No seat licence, no platform fee, no annual commitment, no overage — and credits do not expire. A quiet quarter costs you nothing, and a hiring surge costs the same per candidate as the first one did.