FinHack | Devpost

Flowchart
Architecture
Home Page
ElevenLabs Voice
Personalized Sales and Search Agent
Agent that lives in women's context and prioritizes needs
Personalized Emails for Updates & Reminders | Gemini Veo 2
Agent that lives in women's context and prioritizes needs
Gemini Veo Model, for Duolingo-like Reflection
Snowflake forecasting model
Overspend risk flag (forecast vs last-7-day baseline) and Most uncertain forecasts
Overspend risk Heat Map
Vultr finhack-llm-1 Server Monitor
Vultr finhack-llm-1, Alma Linux 8 x64
USER view: Personalized Advisor with Real-time Updates & Comparisons
EMPLOYEE view: Interactive Employee System monitors users data
EMPLOYEE view: Real-time Supabase Alert
EMPLOYEE view: Customer Service Support + AI Writer
EMPLOYEE view: Chat Bot
EMPLOYEE view: Real-time Supabase Monitor Over Time
Our Supabase

Inspiration

Managing money is stressful, and most budgeting apps add more work: manual logging, guilt-heavy alerts, and features that don’t reflect women’s real lives (income gaps, caregiving, household planning). We wanted a tool that feels supportive, fast, and actually sustainable—something you can use daily without effort, and that helps you save with friends or family in a transparent way.

Our Solution

We built a women-first finance assistant powered by AI agents that live in your spending context. You can log expenses by voice in seconds, get friendly coaching instead of shame, compare prices to find better-value alternatives, and track shared saving goals with clear accountability—turning everyday spending into long-term financial freedom.

Our key features

1) Women-first money support Built around women’s realities — income gaps, care responsibilities, and household planning — with supportive guidance instead of guilt.

2) Smarter spending decisions (women-first shopping + price intelligence) We compare past purchases, surface better-value alternatives (including women-owned / women-led options), and clearly explain why each recommendation shows up.

3) Save together, stay accountable Shared wallets and group goals with transparent contributions and progress — so saving with roommates/partners/friends is simple and clear.

4) Zero-friction tracking Users can just say what they spent. We automatically extract the amount/category/date and update dashboards instantly.

Tech Stack

Layer	Technologies	Key Use
AI + Intelligence	Gemini, Claude 3.7 Sonnet, Snowflake ML Forecast, ElevenLabs	LLM assistants (employee + support), forecasting spend risk, and voice-to-text for faster ops workflows.
Data + Storage	Snowflake, Supabase, MongoDB, pandas, NumPy	Store analytics + transactions, run queries, and transform data into KPIs, cohorts, alerts, and charts.
App + Product	Streamlit, Node.js, Express.js, TypeScript, Figma	Company View dashboard + backend APIs, plus UI/UX design and page flows for a clean demo experience.
Infra + Integrations	Vultr, Docker, Terraform, Solana, SerpApi, Resend, AWS	Deploy + scale services, containers + IaC, blockchain audit trail prototype, external search enrichment, and email notifications.

Prize fit

Prize category	Why we’re a nice fit
MLH: Best Use of Gemini API	Gemini is the “brain” of our app: an AI coach that lives in your spending context, spots patterns, prioritizes goals vs needs, and sends personalized next steps (not generic advice). It also powers our Employee Assistant to explain spikes/risk in plain English for fast ops decisions.
MLH: Best Use of Snowflake API	We used Snowflake ML Forecast to generate 7-day spend predictions + uncertainty bands, then turned that into risk detection (NORMAL / MED / HIGH) vs a baseline. It’s fast, warehouse-native ML with outputs that plug straight into the dashboard.
MLH: Best Use of Vultr	We containerized and deployed services on Vultr so the demo isn’t “works on my laptop.” Real endpoints, real health checks, and a setup that can scale into production monitoring.
MLH: Best Use of Solana	We prototyped a tamper-resistant audit trail on Solana Devnet: key events like “risk flagged,” confirmations, and important alerts can be written on-chain for trust + transparency in fintech workflows.
MLH: Best Use of ElevenLabs	We added voice as a low-friction interface: voice-to-text capture for transactions and hands-free prompts for assistants—so users can log spending instantly and get spoken, supportive coaching back.
Best First Time Hack	3/4 of us are first-time hackers, and we still shipped end-to-end: ingestion → forecasting → risk flags → AI coaching → company dashboard → deployment. We focused on a clean demo that runs reliably.
Capital One: Best Financial Hack	We reimagine banking as proactive and personalized: predict spend, detect risk early, reinforce better habits, and convert raw transactions into clear actions. It’s not just analytics—it’s an always-on finance coach + ops view.
Best UI/UX Design Hack	We built a clean “company view” dashboard with an executive overview, trends, heatmaps, and assistants—everything updates in real time, and the UI makes complex finance signals feel simple and readable.
Best Hack to Support Women	Women-first by design: low-friction capture, supportive coaching (no guilt), smarter purchase discovery, and shared saving transparency. The system is built to reduce money stress and help users build consistent saving habits over time.