Honesti

Consented interview-integrity monitoring

Review the cues. Not a verdict.

Honesti watches a remote interview for signs of outside help and records each one as evidence a person can read — two independent readings, never collapsed into a single score that decides for them.

Create a recruiter accountSign in
One cue, as Honesti records it
Remote-control tool active — AnyDesk
High severityDefinitiveverified publisher

Severity is how much a cue could matter. Strength is how firmly the signal supports it. Honesti keeps them apart so the person reviewing decides — the tool never folds them into one number.

Two readings, kept apart

Every signal carries two independent measures. Collapsing them into one risk score hides the reasoning; keeping them separate makes a cue reviewable.

Severity — impact

How much it could matter

Shown as a restrained chip, from info through critical. It comes from the rule that fired — not from alarm color or a running tally.

Strength — confidence

How sure the signal is

Shown as filled segments — weak, strong, or definitive. A verified software publisher reads stronger than a name that merely looks like a match.

What it watches for

The patterns that usually mean outside help during a remote interview — read from the machine’s own state, not its contents.

AI-assist apps — on-screen answer tools and assistants running alongside the call.

Remote-control tools — screen-sharing or takeover software such as TeamViewer or AnyDesk.

Hidden overlays — windows kept always-on-top, off-screen, or excluded from screen capture.

Extra displays — additional monitors connected during the session.

How a session runs

Four steps, in order — the candidate is never monitored without seeing what that means first.

  1. A recruiter creates a session and sends the candidate a pairing code.
  2. The candidate reviews exactly what is observed, agrees, and opens the agent.
  3. Detections stream to a live timeline as the interview happens.
  4. A reviewer reads each cue on both axes — and decides.

Built on consent

Candidates see precisely what the agent observes before anything starts. Monitoring runs only while the agent is open, only during the interview, and can be stopped any time by closing it.

It observes
  • The names of applications and processes running on the computer.
  • The titles of visible windows and browser tabs (not their contents).
  • Whether a screen-sharing or remote-control tool is active.
  • How many displays are connected.
It does not collect
  • The contents of the screen, documents, or messages.
  • Keystrokes, passwords, clipboard, or browsing history.
  • The camera or microphone.
  • Anything on other devices, or outside the interview.

What it observes are review cues for a person — not automated judgements about anyone.