Amoris

Home page
Add recipient
Chat with bot
Dashboard
Savings Screen

Inspiration

We've all panic-bought a gift. It's 11pm, the occasion is tomorrow, and you're doom-scrolling Amazon hoping something jumps out. We wanted to fix that. Not with another wishlist app, but something that actually thinks with you.

What it does

Amoris is part gift advisor, part savings buddy. You chat with it like you'd text a friend, tell it who you're shopping for, what they're like, what they're into. It asks the right questions, gets a feel for the person, then pulls real buyable gifts from Amazon in real time, ranked by how well they actually fit your budget and their personality.

Found something you love? Save it. Amoris tracks everything you've saved for each person and helps you put money aside for it with a countdown to the occasion, daily savings targets, and milestones that keep you going.

How we built it

Two CS students, one deadline, zero sleep. We built the whole thing in under two weeks for the H0 hackathon.

The stack is Next.js on Vercel, DynamoDB on AWS, Groq's Llama 3.3 for the conversational brain, and RapidAPI for real-time Amazon product data. The AI doesn't just suggest gifts it builds targeted search queries from the conversation, runs them in parallel by interest category, scores every result by price fit, star rating, and popularity, and surfaces the best matches. The savings tracker lives on the recipient profile, so everything stays connected.

Challenges we ran into

Getting the AI to reliably output structured search blocks mid-conversation without announcing them or leaking raw JSON took serious prompt engineering. Multi-query parallel search with deduplication and scoring was trickier than expected. DynamoDB GSIs decided to take way too much time to provision at the worst possible moment.

Accomplishments that we're proud of

The scoring algorithm actually works results feel relevant, not random. The conversational flow feels natural; it doesn't feel like filling out a form. And the full loop from "I need a gift" to "I'm saving for it" actually closes, end to end, in a single app.

What we learned

A well-crafted system prompt is worth more than a fine-tuned model. That DynamoDB will humble you. That the best products you build are the ones that solve a problem you've personally felt. And that two people building the same app from different machines at the same time requires more communication than either of you expect.

What's next for Amoris

Group gifting perhaps, split a savings goal with friends. A memory vault that remembers what everyone in your life likes across occasions. Smarter delivery filtering so you only see gifts that can actually arrive in time. And native mobile so you can use it the moment inspiration strikes.