hSavvy

• 

  Main Page

• 

  Dashboard 1 (Purchase Trends, Reimbursement Amount)

• 

  Dashboard 2 (User Insights)

• 

  Dashboard 3 (AI Generated Insights)

• 

  Transactions

• 

  Drug Pricing (CostPlusDrugs)

• 

  Medical Procedure Lookup (MDSave)

• 

  Portfolio Breakdown

• 

  HSA Reimbursements Item Selection

• 

  HSA Reimbursement Form Generation

• 

  Business HSA Insights Dashboard (Spending, Top Merchants, Monthly Trend, AI Insight Summary)

Inspiration 💡
We kept seeing friends juggle HSAs with spreadsheet chaos, lost receipts, and constant "is this eligible?" anxiety. That pain became our north star: build a benefits co-pilot that quietly ingests the data, explains it in plain English, and nudges you toward smarter decisions.

What It Does 🚀

• Async FastAPI backend ingests Knot-linked commerce data plus manual uploads.
  
• Deterministic analytics: eligibility screening, seasonal spikes, recurring purchases, personalized portfolio allocations.
  
• Gemini-assisted storytelling turns the raw numbers into TL;DRs, key insights, and enterprise rollups.
  
• Action layer: reimbursement PDF generation, MDsave procedure search, CostPlus drug lookup, so insights lead directly to savings.

How We Built It 🛠️

• Stack: Python, FastAPI, async SQLAlchemy with SQLite for hackathon speed, Docker-ready for production.
  
• Security: JWT + bcrypt auth, shared get_current_user dependency for least-privilege access.
  
• Analytics + AI: services/insights computes aggregates, Gemini models add human-friendly summaries.
  
• External data: MDSave + CostPlus APIs for price transparency, Knot client for real-world commerce feeds.
  
• Architecture: Modular routers (/usersdata, /trends, /enterprise, /procedures, /portfolio, etc.) keep the surface area clean and composable.

Challenges We Ran Into 😅

• Choosing a stack under time pressure: FastAPI/Vite vs. going full TypeScript end-to-end.
  
• Four backend engineers sprinting on a full-stack web app with limited frontend experience.
  
• Half the team had never met in person, so aligning on workflows and communication norms took effort.
  
• Settling on the idea itself. We committed barely 20 minutes before hacking officially began.

Accomplishments We're Proud Of 🏅

• Leveling up on a brand-new frontend toolchain while shipping features.
  
• Discovering, reverse-engineering, and stabilizing public APIs to power the core experience.

What We Learned 📚

• There's a surprisingly untapped intersection of healthcare, HSAs, and personal finance, especially once you add investment analytics.
  
• AI copilots shine when paired with real, trustworthy data pipelines; we learned how to orchestrate both across multiple stacks.

What's Next for hSavvy 🔮

• Roll out secure user data vaults so individuals can upload receipts, EOBs, and doctor notes for automated classification.
  
• Ship role-based enterprise dashboards with anonymized cohort analytics so HR teams can benchmark utilization without touching PHI.
  
• Integrate claims ingestion (ELI/EOB parsing) and bank syncing for fully automated reimbursement workflows.
  
• Layer in smart nudges, e.g., "You're $120 short of this quarter's contribution target" or "Book your PT session now for lower pricing", to turn hSavvy into a proactive coach.

Built With

• bcrypt
• costplusdrugs
• docker
• fastapi
• gemini
• html5
• knotapi
• mdsave
• oauth2
• pydantic
• python
• python-jose
• react
• requests
• sqlalchemy
• sqlite
• tailwind
• typescript
• uvicorn
• vite