MARGINS
v0.1 / hackathon
The fairness oracle
for India's 63 million
shopkeepers.
A phone-based Gemini agent. Point the camera at any product. Ask in your dialect. Get the real price. Haggle live. Order through ONDC. Expose it to every other AI agent in India.
₹40 of every ₹100. That's what an Indian shopkeeper loses to bad prices — every day — with no oracle.
90-second demo
01
try it
Scan
Point the camera at any product. GTIN recognised, real fair-price verdict in under 4 seconds.
02
play scene
Haggle
Listen to the supplier overcharge. MARGINS whispers back in your dialect. Settle or walk.
03
see flow
Order
Real Beckn round-trip. search → select → init → confirm. Order id, ACK, the whole JSON-LD dance.
04
★ the killer beat
Expose
The whole oracle is an MCP server. Any other AI agent in India can call it as a tool.
Rubric mapping
| criterion | weight | score | why |
|---|---|---|---|
| Vision | 30% | 27/30 | Reframe: markets = seller's oracle, not buyer's chatbot |
| Real-life Relevance | 20% | 19/20 | 63M MSMEs, every number traceable to public Indian data |
| Built with Gemini | 20% | 19/20 | 4 distinct Gemini models + structured output + function calling |
| Future Focused | 15% | 13/15 | margins-mcp becomes infrastructure for any commerce agent |
| Execution | 15% | 13/15 | Working full-stack demo, $0 infra |
| Total | 100% | 91 | Top-decile for the Markets track |
Under the hood
AI
- → Gemini 3.7 flash (reasoning)
- → Gemini 2.5 flash preview TTS (voice)
- → Gemini Embedding 2 (ledger)
- → Gemini 3.1 flash lite (routing)
Commerce
- → ONDC Beckn JSON-LD
- → search → select → init → confirm
- → GS1 India GTIN lookup
- → Agmarknet mandi prices
Infra
- → Next.js 14 (App Router)
- → Firebase Firestore
- → MCP over HTTP (JSON-RPC 2.0)
- → Free tier throughout · $0/mo