Why Zero-Friction Beats 'Secure Environment' Proctoring for Recruiters
Most 'zero friction' interview proctoring tools quietly ask for camera, screen, and tab access, or swap in their own meeting link. Here's what genuine zero friction actually requires, and why it catches more, not less.
"Zero friction" has become the phrase every interview-proctoring vendor reaches for, and most of them don't actually mean it the way it sounds. There's a real difference between "easy to turn on" and "genuinely nothing changes about the meeting" — and that difference is exactly where a candidate notices something is off.
What "zero friction" is supposed to mean
The honest version of the claim is simple: the interview looks and works exactly like it would have without the tool. Same meeting link, same platform, nothing to install, nothing extra to grant access to, no different experience for the candidate joining.
What it often actually means
Sherlock AI is a good example of the gap between the two, and it's worth being specific rather than vague about it, since the whole point is precision. Sherlock markets itself around ease of setup — connect a calendar, toggle it on — but its own product description asks for camera, microphone, screen, and active-browser-tab access from the candidate, and it works by generating its own "secure meeting" link that replaces the interview's actual Zoom, Google Meet, or Teams link rather than joining that link directly. That's not nothing: the candidate isn't joining the meeting they were originally invited to, they're joining a different, Sherlock-hosted version of it, with broader device permissions granted along the way.
None of that makes Sherlock a bad-faith product — screen and tab monitoring is a legitimate detection approach, and being upfront about what it requires would be a perfectly fine way to market it. But calling that "zero friction" stretches the term past what it actually means for the person on the other end of the call.
What genuine zero friction actually requires
A useful bar to hold any "zero friction" claim to:
- The candidate joins the actual meeting link — not a substitute "secure" version of it.
- No extra permissions beyond what any meeting bot already needs — audio, to build a transcript. Not the screen, not other browser tabs, not the camera.
- Nothing to install, and nothing to toggle on a per-candidate basis that changes what the candidate experiences.
- It works the same way for every role and every meeting, not a setup that has to be configured differently depending on what's being interviewed for.
Comparing the actual requirements
| Lockdown proctoring | Secure-environment tools (e.g. Sherlock-style) | Genuinely zero-friction | |
|---|---|---|---|
| Meeting link | Often a dedicated portal | A generated "secure" substitute link | The real Zoom/Meet/Teams link, unchanged |
| Candidate permissions requested | Browser lockdown, sometimes an install | Camera, microphone, screen, active tabs | Audio only, as a normal participant |
| Feels different to the candidate | Yes, obviously | Yes, even if marketed as low-friction | No — looks like any other meeting |
| Works for any role, unmodified | Depends on setup | Depends on setup | Yes |
For every approach side by side, and which tools use each, see our guide to AI interview proctoring software.
Why "any role, anytime, anywhere" actually matters
A lot of proctoring tools are built with a specific interview type in mind — a technical screen, a timed assessment — and need reconfiguring, or simply don't apply cleanly, outside that context. Genuine zero friction means the same thing happens regardless: a behavioral screen, a technical round, a culture-fit conversation, a call booked five minutes ago with no setup done in advance. If a tool needs to be told what kind of interview it's protecting, or needs the meeting itself replaced with a different link, it isn't really "any role, anytime, anywhere" — it's "this specific kind of interview, if you set it up correctly beforehand."
A couple of common questions
Is device/screen monitoring actually more accurate than a zero-friction approach? Not necessarily — it depends what it's checking for. Screen monitoring catches a copilot running visibly on the same device being watched; it has no visibility into a second device, a phone, or a separate machine, which is exactly where a lot of stealth copilots run specifically to avoid this kind of monitoring. Reading the interview's own transcript and audio doesn't have that blind spot, because it isn't trying to watch a device at all.
Doesn't more device access mean more evidence if something looks wrong? In principle, yes — but only for what's happening on the monitored device specifically. A candidate can grant every permission requested and still run a copilot on a phone sitting outside the camera's view; the extra access doesn't close that gap, it just adds friction and data collection that a transcript-based approach doesn't need in the first place.
Is Candit really zero friction, or is that also just marketing? The bot joins the interview's actual Zoom, Google Meet, Microsoft Teams, Slack or Webex link — the same one that was already going to be used — with no substitute link generated and no camera, screen, or tab access requested. As far as we've seen, that makes it the only AI Detectability option on the market that's genuinely zero friction: any role, anytime, anywhere, in the meeting that was already happening anyway.
Where this shows up in practice
Candit's AI Detectability runs by default on every session without changing anything about how the interview happens. No secure link to swap in, no permissions beyond the ones any notetaking bot already needs, and nothing different for the candidate whether the role is a five-minute screen booked that morning or a scheduled technical round. It reads the live transcript and audio, checks it against reverse-engineered builds of real stealth copilots, and flags an AI-assisted answer the moment it crosses the threshold — with the evidence behind it, in the meeting that was always going to happen. Counter AI and the AI use report add to that without adding friction: the follow-up suggestion shows only to the interviewer, and the report arrives after the call.
First session free. No card required.
