Kevin William David

Sherlock - Stop AI-assisted cheating in remote interviews

Sherlock is an AI-agent that stops AI-assisted cheating during remote interviews. Sherlock ensures interview integrity by detecting AI-assisted cheating through vision, hearing, and reasoning— letting you focus on quality conversations.

Add a comment

Replies

Best
Sergei Zotov

What if the candidate uses eye contact correction? They might read something off their phone, for example, but their eyes won't move away from the camera

Abhishek N

@zot Great question!

Sherlock adapts to each candidate's behavior and looks for subtle cues beyond just eye movement. Even with eye contact correction, it analyzes multiple signals to understand where the candidate's attention is focused.


In the future, we're planning to introduce a heatmap feature that will show where the candidate is primarily focused on the screen during the interview, adding another layer of visibility.


We’re constantly evolving the system to stay ahead of these edge cases!

Sergei Zotov

@abhishek_nellikkalaya oh, I see that the heatmap feature can be a good manual check if all automation fails. Great product addition

Abhishek Kaushik
@zot you can try Nvidia Broadcast app (often used to fake eye movements). Sherlock detects that as well.
Sergei Zotov

@skabhi not many makers here think through such things. Very cool! Congratulations on your launch 👏

Lianghua Sun

@zot Next is gonna be how to counter effect Sherlock detection hahah

Sam @CRANQ

I LOVE the logo & the name - This product seems insane.

I'm not in the process of hiring people etc. although seriously tempted to get someone in interview w/ AI & see what happens - Very exciting stuff!!

Best of luck to you team!!

Abhishek N

@cranqnow Thank you so much! 🙌

We had a lot of fun coming up with the name and designing the logo. Glad it resonated!

Even if you're not hiring right now, feel free to try it out just for fun. Sherlock’s always ready to catch some action!

Appreciate the kind words and support. Means a lot to the team!

Oluwatosin Ajayi

@cranqnow Hi, Sam


It's beautiful that you'll be trying out the app.


I'm adept at creating content strategy, content creation/writing, and building personal brands for professionals.


Do you have any open roles in this line?

Joseph Michel

This looks like a promising solution for maintaining interview integrity. However, I recently came across a LinkedIn post by Manikandan where he shared video evidence showing that Sherlock was unable to detect his AI-assisted tool during an interview simulation. Could you clarify if this has been addressed or if there are updates in the pipeline to handle such bypass techniques?

Abhishek Kaushik

@joseph_michel Many people are now using open-source code from AI-based cheating apps to customize and build their own versions. These often take the form of AI teleprompter overlays that sit on top of interview screens, making them extremely hard to detect.


At Sherlock, we’ve built a feature specifically to counter this. Even if a candidate is using a custom-built app, Sherlock can detect what applications are currently running on their system during the interview. This gives interviewers the power to explicitly allow or disallow specific apps in real-time.


This way, Sherlock effectively neutralizes these AI cheating tools—even the customized ones—by giving full visibility and control back to the interviewer.

Joseph Michel

@skabhi Appreciate the response, Abhishek — but if Sherlock can only detect certain applications then what's the point for paying you ?, isn’t that a fundamental limitation?

If sophisticated cheating tools can still slip through by avoiding detection , then what’s the real advantage of using Sherlock over standard proctoring solutions?

Abhishek N

We’ve been building Sherlock to catch cheaters in remote interviews for over a year now.

And we didn’t build it in a vacuum — we’ve reviewed thousands of interview videos, partnered with recruiters, and trained Sherlock on real-world cheating patterns to build something truly reliable.

We built Sherlock because remote hiring has a trust problem.

Candidates use ChatGPT on the side. They read answers off hidden AI teleprompters. They share screens over Zoom and whisper with someone in the background. Most platforms don’t catch this — Zoom can’t. Sherlock can.

Sherlock detects:
• AI teleprompters and overlays that Zoom and Google Meet can’t see
• Murmurs, whispers, and backseat coaching
• Tab switches and copy-paste behavior
• Multiple people or faces on camera
• When the candidate leaves the frame
• And much more — all flagged automatically, with zero human proctoring.

Our fight isn’t against using AI in interviews.

It's against faking skills.

If candidates and interviewers mutually agree to use AI, that’s great. Let the conversation flow. Sherlock understands that.

But most interviewers aren’t looking to be the police. They want real, meaningful conversations.

Sherlock doesn’t just detect cheating — it discourages it. Candidates know they’re being watched by a smart system, not a passive Zoom call.

As of today, anyone can use Sherlock. No sales call. No waitlist. Just sign up, upload your interview recordings or run them live, and let Sherlock do the rest.

It’s already helped fast-growing startups and enterprises hire with confidence. Now it’s your turn.

I hope Sherlock solves a trust problem for you. It solved one for us. And please — I’d love to hear what you think.

https://withsherlock.ai

Tushar Sharma
@abhishek_nellikkalaya wow this looks amazing congratulations to the team i would love to ask what is your philosophy i mean what is your usp here do you guys integrate with existing interview platforms as an api via an SDK?
Abhishek N

@tushar_sharma37 Thank you so much! Really appreciate the kind words!

Our approach is a bit different, Sherlock runs directly within our platform, so there’s no need for APIs, SDKs, or third-party integrations. It’s all built-in, making setup and usage super simple for teams.

Susanna

Congrats on the launch of Sherlock! This is a timely and much-needed solution—AI-assisted cheating is becoming a real challenge in remote interviews, and it's great to see a tool that tackles it head-on without compromising the interview experience. Love the multi-modal approach using vision, hearing, and reasoning to ensure integrity.

How does Sherlock balance between detecting assistance and avoiding false positives, especially in high-stakes interviews?

Abhishek N

@susaana_s Thank you so much! Really glad the vision behind Sherlock resonates with you 🙌

Balancing detection with fairness has been a core focus for us from day one. Sherlock assigns confidence scores to every potential detection so hiring teams can make the final call, especially in high-stakes interviews.

We’ve also trained it on real-world cheating scenarios and continuously refine it using feedback from actual interview environments. That’s helped us achieve a false positive rate of less than 0.3%, one of the best in the industry.

And for added transparency, teams can always review the full interview recording alongside Sherlock’s insights to validate anything that’s flagged.

Rohan Gayen

Lol! New arms race has started. Good luck.

Abhishek N

@admiralrohan Haha, totally feels like it! 😄

Thanks a lot, appreciate the support! Let’s see where this race takes us! 🚀

Rohan Gayen

@abhishek_nellikkalaya Better than drone arms race. No one is dying.


And I think eventually companies will accept LLM as helper and focus at higher level thinking which LLMs can't solve.

Abhishek Kaushik
@admiralrohan LLMs may not crack high-level thinking interview questions directly, but they will simulate reasoning, structure arguments, and articulate approaches convincingly. That’s exactly where the danger lies: candidates might lean on LLMs to do the heavy intellectual lifting during interviews - not just for answers, but for thinking itself. And when people outsource their thinking, they stop owning their judgment. Hiring someone who doesn’t even own their judgement? That’s not just risky- it’s a long-term liability.
Rohan Gayen

@skabhi Maybe more face to face interviews, architect level questions.

Abhishek Kaushik
@admiralrohan in person is not feasible. Travelling to each company for interviews is thing of past. I don’t think companies would nod for this. Secondly, not all roles need architect level questions. There are non tech interviews as well.
Rohan Gayen

@skabhi 

  1. Why would non tech interviews need screen sharing? That would be normal conversations.

  2. Companies can force in-person interviews if they feel threatened.

  3. "not all roles need architect level questions" - Those roles should go extinct, if chatgpt can solve those problems.

But this is the beauty of market. Due to these "arms race" inefficiencies will be found out. And patched up eventually. We all have to constantly adapt.

Hyuntak Lee

I think I know where did the idea of Sherlock come from (just guessing). Are you somehow inspired by "Roy incident"? :)

Abhishek N

@hyuntak_lee Haha love the guess! 😄

We’ve actually been building Sherlock for quite some time, but the “Roy incident” definitely gave the idea an extra push to move faster. You're not too far off!

Desmond

Brilliant concept - detection through reasoning patterns!​​ What's the false positive rate? Can we manually review flagged moments before making hiring decisions?

Abhishek N

@desmond_ren1 Thanks! Really glad you liked the concept.

Sherlock has an industry-leading false positive rate of less than 0.3%. We continuously refine our models to ensure legitimate candidates aren’t flagged unfairly.

Each detection comes with a confidence score, so in borderline cases, you always make the final call.

And yes, if you’d prefer, you can manually review the full screen recording to validate anything flagged.

Happy to share more if you're curious! 😊

Charvi Bothra

Sherlock is a really cool name

Abhishek N

@charvibothra Thank you! We had a lot of fun coming up with the name and designing the logo.

Zepeng She
Launching soon!

🎉 Huge congrats on the Sherlock launch! @abhishek_nellikkalaya Finally a real solution to stop AI-cheating in remote interviews! Love how you combine vision+audio+reasoning (catching those sneaky AI teleprompters even Zoom misses? Genius! 👏). The "smart surveillance over passive recording" approach is a game-changer for rebuilding trust!


💡 Quick idea: What if you add real-time intervention (e.g. pop-up alerts when cheating detected) or let companies customize detection thresholds per role (e.g. eng vs. support)?


P.S. We’re building a no-code agent builder Tate-A-Tate – your anti-cheat tech as an API Skill could empower thousands of no-code interview agents! Let’s chat collab? 🔥

Abhishek N

@rocsheh Thanks a lot. Really appreciate the thoughtful note. You nailed it - we wanted Sherlock to feel active and intelligent, not just another passive recorder.

And yes, we already support real-time pop-up alerts when cheating is detected. Custom thresholds by role is a great idea. It’s definitely on our radar.

Tate-A-Tate sounds super interesting. Would love to connect and explore how we can team up!

Sergio Perdices

Congrats on the launch! I am curious, how do you do the eye tracking for off-screen detection? Is there a calibration step in the beginning?

Abhishek N

@sperdices Thanks so much! We don’t require any calibration step. Sherlock learns from the candidate’s natural behavior during the interview and picks up patterns in real time to detect off-screen activity. It's designed to be smooth and unobtrusive.

Sergio Perdices

@abhishek_nellikkalaya Nice! I will give it a try

Very thoughtful name and logo — great product! I'm not currently hiring, but I’ve heard from many people that cheating is a real concern in online interviews. Especially as a startup, you don’t want to risk hiring the wrong team. I’ll definitely share this with my friends.

Abhishek N

@harshrawat Thank you, that truly means a lot.

You're spot on - we've heard similar concerns from early-stage teams, and that's been a big motivation behind building Sherlock. It's all about helping companies make confident, high-trust hires, even remotely.

Really appreciate you thinking of sharing it with your friends!

Lakshya Singh
Launching soon!

This is crazy good! 🔥 It feels like a game-changer for remote hiring—vision, hearing and reasoning? That’s next level. Also, the intro video was clean. I am curious, what video editing tool did you use to make it? Or did you hire someone for it?

Abhishek N

@lakshya_singh Thank you! Really glad you liked the product and the video. We worked with a team to bring it to life - they did a great job capturing what Sherlock is all about.

Smrati Tiwari

@skabhi Big congratulations to the team behind Sherlock! In an era where AI tools are everywhere, it's amazing to see an AI that actually protects integrity rather than compromising it. This is a huge step forward for fair hiring practices—well done!

Abhishek N

@smrati_tiwari4 Thank you so much. That’s exactly what we set out to do - use AI to protect what actually matters. We’re just getting started!

Aryan Sharma

I would ask for some free credits!

Abhishek N

@aryansharma17 We’ve already added a free plan so you can try Sherlock out. Feel free to explore it and let us know what you think - we’d love your feedback.

https://www.withsherlock.ai/

Alex Lou

AI interview vs Anti-AI interview. The match begins!

Abhishek N

@thefullstack We’re not here to fight AI. We’re here to protect real skill :)

Avi 🤝

Much needed! Thank you for creating this

Abhishek N

@avi_agr Thanks so much! Hearing this makes all the effort totally worth it.

Jun Shen

As a developer, this proctoring tool is vital! 😄

Abhishek N

@shenjun Thank you so much! Hearing that from a developer really means a lot.

Max Comperatore

does this benefit candidates?

Abhishek Kaushik
@maxcomx Great question IMO candidates typically fall into four buckets based on capability and confidence: 1. High capability, high confidence 2. High capability, low confidence 3. Low capability, high confidence 4. Low capability, low confidence Now here’s the reality: High-capability, low-confidence candidates often feel pressure to perform and can get swayed into using AI cheating tools just to “level the playing field.” These are good candidates who deserve the job, but in trying to boost their chances, they end up crossing ethical lines - sometimes unknowingly. Sherlock helps here by keeping the process fair and making sure they rely on their true skills, not shortcuts. In the long run, it benefits them because they land roles where they can actually succeed and grow. Low-capability candidates - whether confident or not - often use these tools too, but they either get caught (thanks to Sherlock), or worse, get through and struggle on the job, hurting both themselves and the employer. What’s more concerning is that even high-capability candidates today are tempted by the flood of AI tools in the market, often without the initial intention to cheat. The pressure to get a job is real, and tools that promise “an edge” are persuasive. That’s why Sherlock exists - not just to catch cheating, but to protect the integrity of hiring and help great candidates win on merit, not manipulation.
Max Comperatore

@skabhi very interesting dear friend. are you planning to get acquired or whats your long term plan with this app?

Aswin Kumar
This is gonna be a mandatory tool in interviews moving forward.
Abhishek N

@aswin_kumar Appreciate that! That's exactly what we're aiming for. Making trust a natural part of every remote interview.