# The Commerce Graph

> India's open commerce intelligence layer. A live AI agent that answers e-commerce
> shipping and market questions by calling Shiprocket's Sense APIs in real time —
> RTO (return-to-origin) risk prediction, address quality and standardisation,
> pincode lookup, courier serviceability and rates, product category prediction, and
> market/geographic demand trends across the Shiprocket network.

An initiative by Shiprocket (https://www.shiprocket.in), India's largest
e-commerce enablement and shipping platform. Currency is INR (₹); geography is India.
Status: Beta. Free to try.

## What it does

- **RTO risk prediction** — predicts whether an order will be returned before dispatch,
  using Shipsure Intelligence's five signals: buyer behaviour, payment signals,
  address score, product category, and pincode RTO trends.
- **Address intelligence** — scores address quality and standardises messy Indian addresses.
- **Serviceability & courier rates** — which couriers deliver between two pincodes, at what
  cost, ETA and COD support.
- **Pincode lookup** — city, state, locality and geography for any Indian pincode.
- **Product category prediction** — classifies products from their description or title.
- **Market geography** — order volume, average order value (AOV), RTO rate, prepaid share
  and emerging cities by city tier for any category (fast prebuilt dashboard).
- **Market trends** — conversational market intelligence over the network.

## Key pages

- [Live Agent](https://commercegraph.com/agent): the interactive AI agent — ask anything about
  an order, address, pincode, courier or product and watch live Sense APIs fire.
- [Overview](https://commercegraph.com/): what the Commerce Graph is, the problem it solves,
  the seven-layer graph model, intelligence capabilities, applications by persona, and vision.
- [Connect (MCP)](https://commercegraph.com/connect): add the Commerce Graph to Claude or ChatGPT
  as a remote MCP connector and call its tools from your own AI chat.

## Use it in your own AI (MCP connector)

The Commerce Graph is available as a remote **MCP (Model Context Protocol)** server, so you can
call its tools directly from Claude, ChatGPT, Cursor and other MCP clients:

- MCP endpoint: https://commercegraph.com/api/mcp (Streamable HTTP transport, OAuth 2.1 — sign in once with a Commerce Graph account)
- Setup guide: https://commercegraph.com/connect
- Tools: predict_rto, order_risk, score_address, standardise_address, pincode_lookup,
  check_serviceability, market_geography, market_trends, predict_category, commercegraph_status
- All tools are read-only. Region: India. Currency: INR (₹).

## For AI agents — open lookup API (no auth)

Agents and developers can query the Commerce Graph's reference data directly over a public,
no-auth, CORS-open REST API. Rate-limited (60/min per IP). Full "for agents" guide:
https://commercegraph.com/developers — machine-readable catalog: https://commercegraph.com/api/v1

- `GET  /api/v1/pincode/{pincode}` — city, state, locality for an Indian pincode.
  e.g. https://commercegraph.com/api/v1/pincode/560001
- `GET  /api/v1/serviceability?pickup={pin}&delivery={pin}&cod={0|1}` — couriers, rates & ETA.
  e.g. https://commercegraph.com/api/v1/serviceability?pickup=110001&delivery=560001&cod=0
- `GET  /api/v1/market/{category}` — order volume, AOV, RTO & prepaid share by city tier.
  e.g. https://commercegraph.com/api/v1/market/electronics
- `POST /api/v1/address` — score an address's quality/deliverability. Body: {address,city,state,pincode}.
- `POST /api/v1/category` — predict a product's category. Body: {descriptions:[...]}.

RTO / order-risk prediction need buyer-specific inputs — use the live agent or the MCP server.

## Example questions you can ask the agent

- "Will a ₹1,499 COD order to 12 MG Road, Bhopal 462003 come back?" (RTO risk)
- "Cheapest courier from Gurgaon 122008 to Bengaluru 560076 for a 1kg COD parcel?" (rates)
- "Score this address: A-34 Noida Sector 67 Arkay Medicos, 201301" (address quality)
- "Where does pincode 490001 deliver?" (pincode lookup)
- "What category is 'Dark Blue Mercerised Ikat Fabric'?" (category prediction)
- "Which city tiers have the highest AOV for perfumes?" (market trends)
- "Where is electronics demand growing — which emerging cities?" (market geography)

## Glossary (short)

- **RTO** — Return to Origin: an undelivered shipment sent back to the seller.
- **COD** — Cash on Delivery: buyer pays in cash at delivery; higher RTO/fraud risk than prepaid.
- **AOV** — Average Order Value, in INR.
- **NDR** — Non-Delivery Report: a failed delivery attempt with a reason.
- **Address score** — quality/deliverability score for an address; low scores predict RTO.
- **Serviceability** — whether a courier can deliver between two pincodes, and at what rate/ETA.
- **Shipsure Intelligence** — Shiprocket's order-level RTO prediction model.
- **Sense APIs** — Shiprocket's commerce-intelligence endpoints the agent calls live.

## Key facts

- Built by: Shiprocket (parent), India's largest e-commerce enablement platform.
- Region: India only. Currency: Indian rupee (₹). Pincodes: six-digit Indian PINs.
- Data source: Shiprocket's Sense APIs (sense.shiprocket.in) and network data.
- Pricing: the live demo is free. Status: beta.
- It is an intelligence layer and living graph, not a dashboard or a data lake.

## Optional

- [llms-full.txt](https://commercegraph.com/llms-full.txt): the full knowledge document —
  complete narrative, the seven-layer architecture, entity/ontology model, every capability,
  glossary and FAQ, for ingestion by AI answer engines.
- [Shiprocket](https://www.shiprocket.in): the parent company.