Built in 48 hours at our hackathon

DashFlock

Your personal and super cute finance dashboard powered by an on-device LLM. Smart, private, fast, cute — it classifies your expenses automatically so you never lose track on how you wasted your monney.

The flock working together at the hackathon
Penguin mascot
Sheep mascot

What's DashFlock?

A personal finance dashboard that thinks for itself unlike a sheep. Built in 48 hours, powered by a local LLM, and designed to make sense of your spending without snooping on it.

On-Device LLM

All AI inference runs locally on your device. Your financial data never leaves your machine — ever.

100% Private

No cloud, no subscriptions, no data harvesting. Just a penguin and a sheep keeping your secrets safe.

Auto Classification

Import a bank statement and watch DashFlock instantly sort groceries, rent, coffee addiction, and more.

Fuel Cost: 47 Coffees

That's how many cups it took to build this. Worth every sip.

First Place Winner

After 48 hours of intense coding, countless debates, and a few existential crises — we shipped DashFlock. Turns out we were the only team participating, but we'd like to think we would have won anyway.

🏆 1st Place WinnerBuilt in 48 hoursZero data leaks
DashFlock
Your personal finance dashboard powered by an on-device LLM with automatic expense classification.

The Problem

People spend hours manually categorizing expenses. Apps sell your data to advertisers. Cloud solutions create privacy risks. What if your financial AI never left your device?

Our Solution

DashFlock runs a local LLM right on your device to understand and classify transactions automatically. No servers. No data sharing. Just you, your finances, and a very helpful penguin and sheep.

Tech Stack

Next.jsTypeScriptTailwind CSSOllamaWebLLMSQLiteRechartsPythonLangChainBuzzwordCoffeeTeaSnack

The Journey

48 hours of chaos, compressed into the highlights.

Friday 6 PM

The Vision

Bao drew a product roadmap on a napkin. It had 47 features. Nicolas immediately found a bug in the napkin.

Friday 11 PM

Talias Summons the LLM

Talias spent 2 hours configuring the on-device model, then fell asleep on the keyboard. The model actually trained better.

Saturday 2 AM

Luis Disappears

Luis vanished for 45 minutes. He returned with 4 espressos and a working ML pipeline. No one asked questions.

Saturday 4 AM

The Dark Hours

Nicolas found 12 bugs at 4am and fixed them all silently. Anne re-designed the entire UI in the same window. Somehow it worked.

Saturday 3 PM

The Demo Breaks

Five minutes before the practice demo, the LLM classified a grocery receipt as 'Rocket Science'. Talias blamed the training data. The cat was not wrong.

Sunday 6 PM

Champions (uncontested)

We presented DashFlock to a panel of judges. We were also the only team. Bao gave an acceptance speech. It was two minutes long. There was applause.

Meet the Flock

The brilliant (and slightly sleep-deprived) minds behind DashFlock. One cat snuck in, but we kept them.

Bao the Sheep

"Let's add just ONE more feature... and maybe a chart."

Bao the Sheep

Wooly Product Visionary

Nicolas the Penguin

"No bug escapes the Antarctic stare."

Nicolas the Penguin

Ice-Cold Bug Hunter

Talias the Cat

"I taught the model everything it knows. It still can't open a door."

Talias the Cat

AI & LLM Expert

Anne the Donkey

"Pixel-perfect or it doesn't ship."

Anne the Donkey

Fluffy Frontend, Design & QA Lead

Luis the White Cat

"I see patterns everywhere. Especially in espresso foam."

Luis the White Cat

ML Vision & Coffee Expert