Busicash

An AI governance app to evaluate and execute business operations for clients ages 18–25. Gemini analyzes and labels approved purchases as helpful or necessary for users.

Comment

Inspiration

Student entrepreneurs and young founders launching shared ventures (such as campus delivery networks, equipment rental fleets, or print services) face a common barrier: managing capital pool governance fairly. Traditional banking tools lack automated checks, while manual spreadsheets cause co-founder friction over small purchase decisions. We built BusiCash to bring automated AI-driven financial guardrails and multi-sig governance to student teams, giving every partner full visibility and real-time risk assessment before a single dollar is spent.

What it does

BusiCash acts as an intelligent sandbox and multi-sig financial vault for student ventures:

  • AI Purchase Guardrails: Co-founders submit purchase proposals with cost and business justifications. Gemini 2.5 analyzes the request in real time against the venture's active capital pool and operating scope.
  • Automated Compliance & Verdicts: Requests are instantly categorized as Approved (auto-deducted from balance), Requires Co-Founder Vote (sent to multi-sig consensus), or Rejected.
  • Multi-Sig Consensus Voting: When spend thresholds are exceeded or flagged for review, co-founders review Gemini’s analysis and cast atomic votes (Approve/Deny) in a shared ledger.
  • Multi-Venture Dashboard: Student co-founders can manage multiple distinct ventures from a single interface, featuring full ledger tracking and member access management.

How we built it

  • Frontend: Streamlit dashboard built with Python, featuring real-time state management and dynamic venture switching.
  • AI Engine: Google Gemini SDK (google-genai) running tailored prompt pipelines that evaluate spend proposals against venture balance and rationale.
  • Database & Cloud Backend: Google Cloud Firestore in native mode. Uses $O(1)$ slugified document lookups and subcollections (transactions, pending_votes) to support unbounded scale.
  • Atomic Transactions: Deductions from the shared capital pool run inside Firestore transactional blocks (@firestore.transactional) to prevent concurrent race conditions between co-founders.

Challenges we ran into

  • Concurrency & Race Conditions: Ensuring two co-founders approving transactions simultaneously wouldn't overdraw the capital pool. Solved by implementing atomic transactional updates in Firestore.
  • Prompt Precision: Tuning Gemini's evaluation prompts to return strict, deterministic operational verdicts (APPROVED, REQUIRES_COFOUNDER_VOTE, REJECTED) alongside structured human-readable explanations.
  • Schema Evolution: Transitioning from early flat array-based documents to nested Firestore subcollections without breaking active dashboard sessions.

Accomplishments that we're proud of

  • Built a end-to-end multi-sig governance app powered by Google Cloud and Gemini AI in a single hackathon sprint.
  • Designed an atomic, race-condition-proof cloud database architecture using Firestore Admin SDK.
  • Created an intuitive UI that translates complex financial compliance checks into clear AI feedback for student founders.

What we learned

  • Effective design of cloud-native transaction logic using Firestore transactional decorators.
  • Leveraging Gemini for real-time risk mitigation and dynamic rule evaluation in software workflows.

What's next for BusiCash

  • Structured JSON Outputs: Pinning Gemini evaluations to strict Schema JSON modes for granular ledger reporting.
  • AI Smart Agreements: Auto-generating formal partnership contribution agreements based on initial member inputs.
  • Push Notifications: Integrating Webhooks / SMS alerts to notify co-founders instantly when a proposal requires a multi-sig vote.

Built With

Share this project:

Updates