freakypeaky

About the project

Inspiration

Having a crush can turn the internet into a full-blown investigation.

You see their name once, find their Instagram, then somehow you're 47 tabs deep trying to figure out their favorite artist, what they did last summer, and whether that random coffee shop post from three months ago means anything.

We built freakypeaky to save you from the 47 tabs.

Give us a few public URLs and we'll turn your crush's internet presence into a fully fleshed-out profile: their hobbies, interests, projects, favorite things, notable posts, timelines, and whatever else they have casually left lying around on the public internet.

What it does

freakypeaky helps you answer the questions you probably shouldn't have spent three hours trying to answer yourself.

Give it public URLs and it can:

Find their publicly shared hobbies, interests, projects, and favorite things
Connect scattered posts and pages into one suspiciously comprehensive profile
Build timelines from things they've explicitly posted
Surface recurring themes and interests across their online presence
Point out contradictions, missing information, and things the internet simply refuses to tell us
Generate AI-labeled conversation starters, love letters, and messages based only on the evidence
Show the exact sources behind every claim so you can investigate the investigation

It's basically a cheat sheet for your crush.

Privacy and safety

Okay, we're freakypeaky, not actually creepy.

freakypeaky only works with public URLs that you explicitly provide. It does not access private accounts, private messages, passwords, location data, calendars, or hidden schedules.

It also won't figure out where someone is going to be tomorrow, predict their future plans, impersonate them, copy how they write, or send messages on your behalf.

If the internet doesn't clearly support something, freakypeaky says "we have no idea."

Everything comes from what someone has already chosen to put online publicly. We connect the dots, but we don't cross the line.

How we built it

Underneath the questionable premise is a surprisingly serious evidence pipeline.

Discover: accept public URLs or suggest public profile candidates.
Protect: validate sources, filter sensitive information, and keep unsafe content out of generation.
Understand: organize public pages into searchable sources, documents, and evidence chunks.
Connect: generate source-backed profiles, timelines, uncertainty reports, and review-only message suggestions.

Our stack includes Next.js, React, TypeScript, Tailwind CSS, Python, FastAPI, Pydantic, SQLAlchemy, SQLite, and FTS5.

Challenges

The hardest part was making something that feels "wait, this is kind of insane" without actually making it invasive.

Every generated statement needed to stay tied to evidence. We had to distinguish facts from inferences, preserve uncertainty, filter sensitive information, and make sure a few innocent public breadcrumbs didn't magically become a completely made-up biography.

We also had to make the AI messages useful without letting it pretend it knows your crush better than you do.

What we learned

Apparently, there's a very fine line between "I did my research" and "how do you know that?"

Building freakypeaky taught us that the funniest version of an AI product can also be one of the hardest to build responsibly. The product can feel chaotic and ridiculous on the surface while the system underneath is extremely careful about evidence, uncertainty, and privacy.

At the end of the day, freakypeaky isn't here to uncover someone's secret life.

It's here to take the completely public information they've already left scattered across the internet and turn it into the world's most questionable conversation cheat sheet.