Built for remote technical interviews
See the screen your webcam can’t.
The candidate points a phone at their laptop so you can watch that screen live, while AI names the app on screen and flags any AI assistant for your review.
- Candidate consent first
- No face recognition
- Payments by Razorpay
Webcams show faces, not screens.
Remote coding rounds depend on what happens on the candidate’s screen. Today, that’s the one thing you can’t see.
The webcam faces the wrong way
It shows the candidate. The screen they are typing on stays out of view.
AI sits one Alt-Tab away
ChatGPT, Claude, Gemini or Copilot can run in a window right next to the editor.
Interviewers are left guessing
A long pause or a perfect answer could mean anything. Without the screen, you can’t tell.
How it works
Set up in three steps
No installs, no browser extensions. A phone camera on their screen and a dashboard on yours.
Step 1
Share the link
Create an interview and send the candidate its joining link. They don’t need an account or an app.
Step 2
The candidate points their phone at the screen
On their phone or a second device, they give consent and aim the camera at their laptop. Live checks for framing, lighting, focus and stability guide them until the whole screen is visible.
Step 3
Watch live while AI flags what needs a look
You see their screen in your dashboard. At random moments, AI names the app on screen and raises a popup if an AI assistant shows up.
Features
Everything you need to review a remote round
See it live, get a nudge when something needs a look, and keep a clear record for afterwards.
Live monitor
Watch the candidate’s screen through their phone camera, with live cards for screen detection, visibility, lighting, stability and network.
AI screen detection
At random moments 1–10 s apart, an AI vision model names the app on screen. AI assistants raise an “Unusual screen activity” popup.
Screenshot evidence
Each flagged frame is saved as a screenshot alert in private storage, with the app name and time.
AI summary
After the interview, a 100–120 word summary of the alerts. It’s written from alert text only; screenshots are never sent to the model.
One-click recording
Record the session with one click and play it back later, seeking to any moment.
Event timeline
Everything that happened, in order: joined, screen lost, app changed, review recommended.
Try it
Click through a real-looking timeline
This is what an interviewer sees during a session. Pick any event, or open a flagged one to see its screenshot alert.
Event timeline
Frontend Engineer, Round 2
Selected event
Candidate joined
10:02:14
The candidate gave consent and connected their phone camera.
Try it
Open a Review recommended event to see the screenshot alert an interviewer gets.
The report
One page to review the whole interview
When the interview ends, the summary, screenshots, recordings and timeline sit together, so you can decide with the full picture.
AI summary. 100–120 words written from alert text only. Screenshots are never sent to the model.
Screenshots. Every flagged frame with the app name and time, kept in private storage.
Recording. Play the session back and seek straight to the moments that matter.
Timeline. Everything that happened, in order, so the alerts have context.
Interview report
Frontend Engineer, Round 2
18 Nov 2026 · 52 min
AI summary
The screen showed VS Code and a terminal for most of the session. At 10:24 the AI scan named ChatGPT and a screenshot was saved; at 10:41 it named Claude. Both are marked review recommended…
Screenshots
Return the indices of two numbers that add up to target
Here’s an O(n) approach using a hash map:
ChatGPT10:24:05
Return the indices of two numbers that add up to target
Here’s an O(n) approach using a hash map:
Claude10:41:48
Recording
Timeline
- 10:02Candidate joined
- 10:24ChatGPT · Review recommended
- 10:25App changed · VS Code
Trust & fairness
Fair to candidates. Clear for interviewers.
Monitoring only works if both sides trust it. These rules are built into the product, not left to policy.
Consent first
Candidates give consent on the joining page before anything starts. Nothing is captured until they do.
Review, not verdict
Every flag reads “review recommended”, never “cheating”. A person always makes the call.
Private storage
Screenshots and recordings live in private storage, visible only to the interview’s owner.
No face recognition
There’s no face detection or recognition. The camera points at the screen, and that’s what the AI reads.
Summaries from text only
The AI summary is written from alert text. Screenshots are never sent to the model.
Pricing
One simple plan
Every feature, one monthly price. No tiers to compare.
One plan, everything included
Interview Detect Monthly
₹10,000/ month
100 interviews every month · about ₹100 each
Get startedSecure payments by Razorpay · renews monthly · count resets each month
What’s included
- Live view of the candidate’s screen through their phone camera
- Guided calibration for framing, lighting, focus and stability
- AI screen analysis at random moments, 1–10 s apart
- “Unusual screen activity” popups with screenshot alerts
- One-click session recording with playback and seeking
- Event timeline, AI summary and a full report per interview
- Private storage, visible only to the interview’s owner
FAQ
Questions, answered
No. The candidate opens the joining link in a browser on their phone or a second device. There’s no app or extension to install.
Yes. The candidate gives consent on the joining page before anything starts. Nothing is captured until they agree.
No. There’s no face detection or recognition. The camera points at the candidate’s laptop screen, and the AI only looks at what’s on that screen.
No. A flag means something on screen, such as an AI assistant, deserves a human look. Every alert says “review recommended”, and the decision is always yours.
At random moments 1–10 seconds apart, an AI vision model names the application on screen, for example VS Code, Chrome, Terminal, ChatGPT, Claude, Gemini or Copilot. If an AI assistant appears, you get a popup and the frame is saved as a screenshot alert.
No. The post-interview summary is written from alert text only. Screenshots are never sent to the model.
Only the interview’s owner, or an admin. Screenshots and recordings are kept in private storage.
You can’t create new interviews until your monthly count resets. The plan renews monthly, and the count starts again at 100 each month.
Your next remote round, with the whole screen in view.
Create an interview, share the link, and watch the screen live. One plan covers every feature.