• 

  Scan menu

• 

  Scanning product Details

• 

  Product details

• 

  Profile page

• 

  Product History

• 

  signature keys by users

• 

  product history short

• 

  wallet deatils

• 

  confirm payment

• 

  Ai for all micro transactions and Ease od use

• 

  scan + pay per use button

• 

  login page with pera wallet

• 

  login page

• 

  payments reciet

• 

  Logo + Initiative

💡 Inspiration

India loses ₹1.5 lakh crore annually to counterfeit products — adulterated medicines, fake electronics, forged luxury goods. As students at Bangalore Institute of Engineering, we've seen this problem up close. Existing supply-chain verification is either too expensive for small businesses or too centralised to be trustworthy.

Then we discovered the x402 protocol — HTTP 402 Payment Required — and realised something powerful: AI agents can now pay for services autonomously. What if we combined product verification with agentic micropayments? Instead of forcing users into expensive subscriptions, what if AI agents could pay per-use to verify products — buying trust like they buy tools in coding environments?

That's when VeriPass was born: an anti-counterfeit product passport where AI agents pay per-use on Algorand to verify authenticity, trace supply chains, and fight counterfeiting.

---

🛠️ What It Does

VeriPass is an x402 product passport platform on Algorand with three core capabilities:

1. Product Verification — Users scan a QR code to view a product's full chain-of-custody: where it was made, who handled it, how it got to the shelf. Each checkpoint is cryptographically signed.

2. AI Agentic System — 7 specialist AI agents (inventory, passport, market, usage, proof, search, compare) that each charge micropayments via x402. A Research Agent demonstrates machine-to-machine commerce by paying three services in sequence: market data → news → report.

3. x402 Micropayments — Every action (scan, query, report) triggers a real Algorand TestNet transaction. No subscriptions, no credit packs — just pay-per-use.

The cost model per action is:

$$C_{\text{action}} = \begin{cases} 0.002 \text{ ALGO} & \text{product scan} \ 0.003 \text{ ALGO} & \text{agent query} \ 0.005 \text{ ALGO} & \text{comparison report} \end{cases}$$

---

⚙️ How We Built It

Frontend: React 19, Vite, TypeScript, Tailwind CSS — a state-based SPA with 13 screens including a live QR scanner, AI chat, wallet dashboard, and passport viewer. Pera WalletConnect for Algorand integration.

Backend: Hono (Node.js) with 30+ REST API endpoints, SQLite via built-in node:sqlite (zero native deps), HMAC-signed authentication, and ed25519 digital signatures for supply-chain checkpoints.

AI: Google Gemini (gemini-3.6-flash) with tool-calling for each agent. Agents autonomously check wallet balances, execute payments, and return verified data. An agentic loop runs up to $N_{\max} = 6$ iterations per query.

Payments: Native Algorand TestNet transactions via algosdk v3. Each user gets a unique wallet. GoPlausible facilitator handles x402 protocol settlement. Payments are direct wallet-to-wallet transfers — no smart contracts.

The system architecture follows:

$$\text{User} \xrightarrow{\text{scan}} \text{React SPA} \xrightarrow{\text{API}} \text{Hono Backend} \xrightarrow{\text{x402}} \text{Algorand TestNet}$$

$$\text{AI Agent} \xrightarrow{\text{tool call}} \text{Gemini} \xrightarrow{\text{payment}} \text{Algorand} \xrightarrow{\text{proof}} \text{Product Passport}$$

---

🧩 Challenges We Ran Into

1. x402 spec was evolving — The protocol changed during the hackathon. We had to adapt our middleware integration multiple times to match the latest @x402/hono API.

2. Wallet management is hard — Each user needs a unique Algorand wallet, but managing mnemonics securely at scale is non-trivial. We solved it with a gitignored wallet store and optional client-side Pera signing.

3. AI tool-calling reliability — Getting Gemini to reliably call tools (wallet check, payment, product lookup) required extensive prompt engineering. The agentic loop sometimes hallucinated tool names — we added strict validation.

4. Cold starts on free tier — Running on Render free tier means ~30s cold starts. SQLite also doesn't handle concurrent writes well. Acceptable for a demo, but would migrate to PostgreSQL in production.

5. Security incident — Wallet mnemonics were accidentally exposed in the initial README push. We caught it, redacted them, and added a gitignore rule. Lesson learned: always audit before pushing.

---

🏆 Accomplishments We're Proud Of

1. Real x402 payments working end-to-end — From QR scan → AI agent → Algorand payment → verified passport, every step is real. Transactions are viewable on Lora TestNet Explorer.

2. 7 AI agents with autonomous payments — Each agent charges micropayments and executes real ALGO transactions without human intervention. The Research Agent pays 3 services in sequence:

$$\text{Research} \xrightarrow{0.003} \text{Market Data} \xrightarrow{0.003} \text{News} \xrightarrow{0.003} \text{Report}$$

1. Zero native dependencies — Using Node.js built-in node:sqlite meant no native compilation, no build failures, and seamless deployment.

2. Full platform — 30+ API endpoints covering auth, verification, AI, payments, digital signatures, and admin dashboard — a complete platform, not a prototype.

3. Made in India — Built entirely by students at Bangalore Institute of Engineering, addressing a uniquely Indian problem.

---

📚 What We Learned

• x402 is real machine-to-machine commerce — Not just a spec, but a working protocol that enables autonomous agent payments. The future of commerce is agents paying agents.

• Algorand is built for micropayments — With finality time:

$$t_{\text{finality}} \approx 3.3\text{s}, \quad f_{\min} = 0.001 \text{ ALGO/txn}$$

Traditional payment gateways charge ₹2–5 per transaction, making micro-verification impossible. Algorand makes it viable.

• Simple architecture wins — No smart contracts, no DeFi complexity — just native ALGO transfers via HTTP headers. Sometimes the simplest solution is the best one.

• AI agents need wallets and budgets — The shift from "humans pay" to "agents pay" requires wallet management, spending limits, and audit trails. We built all three with a spend-policy guard:

$$\text{allowed}(s) = \begin{cases} \text{true} & \text{if } s \leq s_{\max} \ \text{false} & \text{otherwise} \end{cases}$$

where $s$ is cumulative spend and $s_{\max}$ is the budget cap.

• SQLite is surprisingly production-ready — Zero config, single file, built into Node.js. Perfect for hackathon-scale projects.

---

🚀 What's Next for VeriPass

1. Mainnet deployment — Move from Algorand TestNet to Mainnet with real ALGO payments and production wallet management.

2. PostgreSQL migration — Replace SQLite with PostgreSQL for concurrent write support and horizontal scaling.

3. Mobile app — Native React Native app with camera-based QR scanning and Pera Wallet deep linking.

4. Enterprise API — B2B API for manufacturers to register products and issue passports at scale.

5. Multi-chain support — Expand beyond Algorand to support other x402-compatible chains.

6. AI agent marketplace — Let third-party developers deploy their own verification agents on the VeriPass network.

7. Regulatory integration — Connect with Indian government supply-chain digitisation initiatives (Digital India, BIS compliance).

Built With

• agentic
• ai
• algorand
• google-gemini
• made-in-india
• micropay
• node.js
• pera
• qr-code
• react
• sqlite
• supplychain
• x402