Most scams don't look like scams. They look like a job offer, a delivery notice, a message from your bank, a note from someone you already trust. Hunch is the calm second opinion for that exact moment. You paste what you got, and in plain words it tells you what you are being asked to do and the safest next step. It runs on your device, there is no account, and we do not store your messages.
Isn't this just another "scam detector" extension?
Most tools check the link after you click it. Hunch reads the message itself, so it can catch the ones with no link at all: a fake recruiter, a "read me the code" call, a request to move a payment somewhere new. It points at the specific things worth a second look, the pressure, the deadline, the lookalike address, the exact ask, and tells you what it saw and what it would do. Same message in, same answer out, every time. It is not a chatbot that guesses a little differently each time you ask.
Why wouldn't I just paste the message into ChatGPT?
You can, and Hunch even has a button to get an AI second opinion. But ChatGPT only helps once you already suspect something and decide to go ask. Hunch is there the moment the message arrives, before you have decided whether to worry. It runs on your device, answers in about a second, and sends nothing anywhere unless you choose the optional AI step.
A rules engine only catches scams it already knows, so what use is it?
Scams repeat their shape, not their exact words. Hunch matches the shape: urgency, plus a lookalike address, plus a request to hand something over. It does not need to have seen that exact message before. When someone flags a new twist, the pattern can reach everyone quickly. It will not catch everything, and it never pretends to. When it is not sure, it says so, and still gives you a safe next step.
You have no users and no revenue. This is a hobby project, right?
Early stage, yes. A hobby, no. The extension is real and shipping, the engine is built and tested, and the scam library is real. What we will not do is inflate the numbers to sound bigger than we are. If that honesty makes a skeptic raise an eyebrow, good. It is the same honesty the product itself runs on.
Is it free, or what's the catch?
Checking a message is always free. No limit, no account, ever. The only thing you'd ever pay for is real legwork on a high-stakes decision: you're about to send money on a rental, a car, a job, or a wire, and you want us to check whether the specific landlord, seller, or company is real before you do. Most of the time there's nothing to check, and we'll tell you so. (Plus a version for businesses.) That's the paid side.
Do you read, or sell, my messages?
No. Detection runs on your device, inside your browser. Your message is read and scored locally, we do not store it, and we do not send it to us. Nothing is logged. The one exception is the optional AI second opinion: if you turn it on, at most 300 redacted characters of that one message are sent, and only when you ask. You can watch all of this yourself in your browser's network tab.
What if it's wrong? I don't want false alarms.
Hunch shows you things to check, not a final ruling. It is tuned so that ordinary, legitimate messages stay quiet, because a tool that cries wolf is a tool you switch off. When something is worth a look, it tells you what and why, and you decide. When it is not sure, it says so rather than guessing.
Who's behind this? Why trust an unknown startup?
One person, named and reachable, not a faceless company. You do not have to take the privacy on faith: open your browser's network tab and see for yourself what is and isn't sent. The docs say when each claim was last checked and how. Trust should be earned by what you can verify, not asked for up front.
Isn't a scam-detector just fearmongering?
The opposite is the point. Hunch stays quiet on the messages that are fine, and most messages are fine. It exists so you can stop inspecting every message yourself and spend your attention only on the few that earn it. Calm, not noise.
Can't scammers just read your rules and design around them?
The parts that make a scam work cannot be removed without breaking the scam: the false urgency, the request to pay or hand over a code, the impersonated brand, the push to move off the platform. Take those out and there is no scam left. And when someone finds a genuinely new trick, Hunch learns from it quickly.
Doesn't my bank, Google, or the built-in spam filter already do this?
Those rely mostly on lists of threats already known, and they struggle with brand-new lookalike addresses and messages that arrive off-platform, on LinkedIn, on WhatsApp, wherever. Hunch is a second layer on the surfaces they do not cover, and it reads the message itself instead of just matching a sender against a list.
Does it slow down my browser?
No. A normal check is a lightweight matcher that runs in milliseconds, with no heavy model to load and no network call. You will not feel it.
Does it work on mobile?
Not yet. Today Hunch is a desktop extension for Chrome and Chromium, where a lot of email and LinkedIn reading happens. Mobile is not built yet, and we would rather tell you that plainly than pretend otherwise. If mobile is where you need it most, tell us at [email protected], and it moves up the list.
Which languages does it understand?
English first, and it also reads Hebrew, Spanish, and Portuguese. Hebrew matters to us and we put real work into it, and we will be straight with you: it is still improving, and it is strongest in English. If a message in your language gets a weak read, that is on us to keep fixing, not on you.
Do I need an account to use it?
No. No account, no email, no profile to check a message. An account only comes into it on the paid side, when you ask us to check a specific landlord, seller, or company before money moves, or for the business version. The check itself never asks you for anything.
Is my data stored anywhere? Is it GDPR-friendly?
There is very little to store, because the check happens on your device and your message is not kept. That local-first approach is what lines it up with GDPR: process locally, keep only what is truly needed, which here is close to nothing.
What actually happens to a message Hunch flags?
Nothing happens to it without you. Hunch is an advisor, not an enforcer. It shows you what it saw, and you decide what to do. It never deletes, reports, moves, or quarantines a message on its own.
How do I install it?
From the Chrome Web Store: search Hunch, or use the Add to Chrome button on this site. No account, no setup. Paste a message, and you are going.