Inspiration

What it does

How we built it

Challenges we ran int## Inspiration

Over 82% of small and medium-sized enterprises (SMEs) collapse not from lack of profitability, but due to acute cash-flow insolvency and the credit scoring paradox: when seeking institutional working capital, businesses are forced to expose proprietary trade secrets, margins, and client lists. Traditional tools like QuickBooks and Excel are strictly passive logs that record insolvency after cash is gone. We engineered OmniMesh Capital to turn treasury management into an autonomous, edge-first liquidity defense engine and private credit solvency passport.

What it does

OmniMesh Capital functions as an autonomous, edge-native digital CFO and liquidity shield:

  1. Client-Side SHA-256 Merkle Ledger: Transactions form a cryptographically chained ledger evaluated in real time using client memory, featuring out-of-band tamper detection and automated consensus healing.
  2. Monte Carlo Stochastic Runway Simulator: Projects a 120-path Geometric Brownian Motion (GBM) liquidity trajectory across 60-day horizons, isolating 5th-percentile (P05) stress floors.
  3. Dual-AI Autonomous Sentry:
    • Nebius AI Studio (Meta-Llama-3.1-70B): Generates legally calibrated debt collection settlement notices and dynamic early-settlement discounts for overdue receivables.
    • Hugging Face (ProsusAI/finbert): Continuously scores treasury narrative risk and assigns objective delinquency risk ratings.
  4. Zero-Knowledge Solvency Passport: Issues cryptographic proofs via Groth16 zk-SNARKs and Pedersen commitments across 2,418 R1CS constraint gates, validating commercial solvency (Solvency Ratio ≥ 1.25x) to institutional lenders without exposing private balance figures or customer lists.

How we built it

  • Frontend & Runtime: Next.js 14 App Router, TypeScript, and Tailwind CSS styled as a Bloomberg-grade financial operations terminal.
  • Cryptography: Native WebCrypto API for local SHA-256 Merkle linking and Groth16 zk-SNARK constraint verification structures.
  • AI Orchestration: Nebius AI Studio endpoints for 70B model inference and Hugging Face Inference API for FinBERT sentiment classification, backed by deterministic offline fallbacks.
  • Deployment & Tooling: Docker, Docker Compose, and Vercel for zero-config global deployment.

Challenges we ran into

  • Ensuring client-side stochastic Monte Carlo simulations run sub-second calculations over 120 paths without blocking the browser rendering loop.
  • Designing zero-knowledge attestation flows so micro-merchants can prove debt-service ratios purely on edge devices without uploading confidential records to centralized servers.

Accomplishments that we're proud of

  • Delivering a dual-perspective product that bridges rigorous computer science (tamper-proof Merkle chains, ZK circuits) with commercial finance viability (tackling the $5.2 Trillion SME working-capital gap).
  • Seamless offline failover logic ensuring zero 500-errors during live judge testing.

What we learned

Local-first, deterministic cryptographic chains coupled with specialized foundation models can provide robust financial autonomy to small businesses without invasive surveillance.

What's next for OmniMesh Capital

  • On-chain smart contract settlement for decentralized invoice-factoring liquidity pools.
  • Expanding zk-SNARK verification circuits directly to Layer-2 rollups.
  • Mobile edge nodes for micro-merchants in emerging economies.o

Accomplishments that we're proud of

What we learned

What's next for OmniMesh Capital

Built With

  • docker
  • finbert
  • groth16
  • hugging-face
  • llama-3.1-70b
  • nebius-ai
  • next.js
  • tailwindcss
  • typescript
  • webcrypto
  • zk-snarks
Share this project:

Updates

Submission history