Iacob Paștina

Pino - Fact Checker - Fact check highlighted text on the internet with AI

I built Pino to cut through misinformation. This Chrome extension fact-checks claims instantly using AI, providing multiple sources and a truth score. Whether you’re reading news, or researching, Pino helps you separate fact from fiction in seconds.

Add a comment

Replies

Best
Iacob Paștina

Building something from scratch means the world to me.


I’ve always seen devs as modern sorcerers: a few keystrokes and something magical pops out. The first time I prompted an AI felt like stumbling into the App Store circa 2009. Back then I was way too young to think like a creator; I just main‑lined shiny new apps.


I didn’t realize how badly I wanted to make until I opened an AI‑powered IDE. Once it turned fully agentic, my output exploded—call it a 10× buff.


For the record, this entire project was just me, my laptop, and a fortress of pillows (my bed).


I’m emotional because I’ll never again be this clueless and this electrified. It’s the same high I got shooting my first film at 17 or hammering through my first drum gig. A decade later the buzz is back—I built something that actually works. Sure, it’s rough around the edges and using AI can feel like spelunking without a headlamp, but still: it works. And that rules.


Fact‑checking is brutal—sometimes borderline impossible. The world spins faster every day, and we’ve started trusting raw, shaky footage over polished broadcasts. But facts still matter. So I built a system that pings multiple sources, basically taking the journalist’s “triple‑check” ethos and cranking it to eleven.


I named him Pino, after that wooden kid who just wanted to be real. I wanted my Pino to have the same restless curiosity—a protagonist sniffing his way through a tangled world.


If this tool shaves even a few minutes off someone’s search for truth, I’m happy. It doesn’t have to be flawless, just useful.


Next stop: true autonomy and a bolder personality. I want Pino exploring, poking at corners—same as me.


We’re only at page one. Thanks for reading.


—Iacob & Pino

Barnabé DUBUS

Congrats @iacob for your first tool. Very simple and clever idea. Perplexity rocks.

I hope you will catch a lot of users today ;)

Iacob Paștina
@barnabed thanks a lot! It was quite the journey
Jun Shen

Real-time verification adds much-needed credibility! 😄

Shreyans Bhansali

Love the energy here—especially the honesty about being "clueless and electrified." One question: how does Pino decide which sources to trust when it’s triple-checking? That judgment layer feels like the real magic.

Iacob Paștina

@shreyans_assistiv thanks a lot!! under the hood, Pino goes through three quick passes before it will “trust” a source:


  1. Reputation & provenance (≈ 1 ms) – Every URL is run through a lightweight classifier that checks:

    • domain age + history (Common Crawl + Tranco lists)

    • whether it’s an official/institutional TLD (.gov, .edu, etc.)

    • 3-rd-party reliability signals (Media Bias Chart, Google Fact-Check corpus, retracted-paper lists for journals).

  2. Claim–evidence fit (≈ 10 ms) – For the specific sentence being checked Pino embeds both the claim and the candidate passage, measures semantic overlap, and throws out anything that’s mostly commentary or opinion. It also tags the type of source (news, academic, primary data, social) so it can keep the final list diverse.

  3. Cross-source agreement (≈ 30 ms) – It needs corroboration from at least two independent, high-score sources of different types. If Source A makes a factual statement, Pino looks for the same datapoint in Source B. No agreement → no citation.


What comes out the other side is a bundle of 5-10 sources, each with a confidence score. Those numbers drive the “truth-score” UI and the length of Pino’s nose 😅.