Overview
VeriVest Immo is Tunisia's first AI-driven, bank-backed, blockchain-based digital asset wealth platform, built during the AIxCYBER hackathon where it won Best Pitch. An Express.js backend (port 3001) integrates with Hedera Hashgraph via @hashgraph/sdk — HTS (Token Service) for fractional property ownership (1000 fungible tokens per asset) and HCS (Consensus Service) for immutable audit trails. Each property registration generates a SHA-256 document hash, creates an HCS topic for event logging, and issues HTS tokens. A Next.js 14.2 frontend (port 3000) features a French-language investment audit UI with 3D flip cards, animated score gauges, and three risk profiles (Conservative, Balanced, Visionary). n8n webhook workflows handle AI price prediction and portfolio recommendation.
The Problem
Real estate investment in Tunisia requires significant capital, shutting out the majority of potential investors. There's no fractional ownership framework, and the due diligence process is manual, opaque, and takes months. Banks have vetted properties sitting idle because they can't find enough qualified buyers, while small investors have no access to institutional-grade assets. No existing platform combines blockchain-based fractionalization with AI-driven investment analysis.
Solution
Built a blockchain backend (Express.js on port 3001, Mongoose + MongoDB Atlas) with 6 controllers (asset, audit, fraction, risk, transfer, user) and 2 Hedera services (`hederaHCS.js` for Consensus Service topics/messages, `hederaHTS.js` for Token Service creation/issuance). Each property registration generates a unique `ASSET_` ID, computes a SHA-256 document hash for immutability proof, creates an HCS Topic for audit trail logging, and issues 1000 fungible HTS tokens for fractional ownership.
Integrated n8n webhook workflows for AI-powered analysis: (1) an admin property creation webhook that triggers AI price prediction (estimated price in TND, confidence score, min/max range) and posts results back to `/api/price/update`, and (2) an audit flow webhook that receives the user's investment amount and risk profile (conservative/balanced/visionary), runs AI analysis against blockchain-registered properties, and stores portfolio recommendations to MongoDB.
Built the Next.js 14.2 frontend (React 18, Framer Motion, Radix UI Slider, Lucide icons, Tailwind 3.4) with a fully French-language UI. The audit screen features an animated investment slider (€1,000–€100,000), three risk profile cards with shield/trending/zap icons, and a pulsing 'Lancer l'Audit IA' button. Results display as 3D flip cards (600ms CSS rotation) with investment and security score gauges, AI-generated analysis text, and 'Vérifier sur Hedera' links to hashscan.io.
Implemented audit trail retrieval via HCS topic message history — the audit controller queries `getTopicMessages()` for any asset's complete blockchain event log, providing regulators with immutable, timestamped records of every registration, token issuance, transfer, and risk scoring event.
Architecture
Next.js 14.2 frontend (port 3000) proxies requests through Next.js API routes (`/api/admin/properties`, `/api/blockchain/assets`, `/api/blockchain/fractions`, `/api/blockchain/risk`, `/api/audit`) to an Express.js blockchain backend (port 3001). The backend uses Mongoose to connect to MongoDB Atlas and @hashgraph/sdk to interact with Hedera Testnet (account 0.0.7092377). Two n8n webhook endpoints handle AI processing: one for admin price prediction on property creation, one for the user-facing audit flow that generates portfolio recommendations based on risk profile. The frontend uses Framer Motion for staggered card animations and loading state transitions.
Tech Stack
Express.js + @hashgraph/sdk
Blockchain backend on :3001 — 6 controllers, 2 Hedera services (HCS topics/messages, HTS token creation/issuance)
Hedera HTS (Token Service)
1000 fungible tokens per property — fractional ownership with transfer and balance queries via Hedera Testnet
Hedera HCS (Consensus Service)
Immutable audit trail — each asset gets an HCS Topic, every event (registration, issuance, transfer, risk score) logged as a timestamped message
n8n Webhook Workflows
2 AI workflows: property price prediction (estimatedPriceTND, confidence, range) and audit/portfolio recommendation (risk-aware asset allocation)
Next.js 14.2 + React 18
French-language UI — Framer Motion animations, Radix UI Slider, 3D flip cards, animated score gauges, Lucide icons, Tailwind 3.4
MongoDB Atlas + Mongoose
3 collections: assets (property + blockchain IDs), portfoliorecommendations (AI allocations), riskscores (0–1 scale with model version)
Results
Won Best Pitch award at AIxCYBER hackathon
HTS fractional ownership starting from small amounts — SHA-256 hashed for immutability
Every event logged on HCS Topics — verifiable on hashscan.io/testnet
Conservative, Balanced, Visionary — AI tailors portfolio recommendations per profile
Automated price prediction with confidence scores and portfolio optimization
Achievement Gallery
Moments that made it all worth it
Achievement 01
🏆 Best Pitch Award — AIxCYBER Hackathon 2025 at Esprit School of Business
Key Takeaways
Hedera's token service is significantly faster and cheaper than Ethereum for asset tokenization — sub-second finality and predictable ~$0.001 fees. The @hashgraph/sdk made HTS/HCS integration straightforward from Express.js.
n8n's webhook-based architecture was perfect for the hackathon — we could iterate on the AI analysis logic (price prediction confidence scores, risk-profile-based allocation) visually without touching backend code.
The 'Best Pitch' win validated that investors care more about the business model (fractional ownership of Tunisian real estate) than the technology (blockchain). Lead with the problem, not the stack.
Building the entire UI in French (from 'Lancer l'Audit IA' to 'Vérifier sur Hedera') was a deliberate decision for the Tunisian market — localization matters more than technical polish for pitch judges.