DOCS · GETTING STARTED

Build on Aiva

Everything Aiva knows about a store, you can reach in code: a keyed REST API with interactive Swagger docs, a full MCP server that turns her data into tools for Claude, Cursor and any MCP client, and automation recipes for n8n, Zapier and Make. Grab an API key and merchant ID from your merchant dashboard and you’re querying real data in five minutes.

On the packages. The MCP server is published and installable as @getaiva/mcp, and the app scaffolder as @getaiva/create-app. There is no published JavaScript or Python SDK yet, so the examples below call the REST API directly, which is the supported integration path today.

API quickstart

Send a customer message to Aiva and get her reply - with confidence score and an escalation flag, so your app knows when a human should step in:

POST /V1/CONVERSATIONS
const res = await fetch("https://api.getaiva.ai/api/aiva/message", {
  method: "POST",
  headers: {
    "X-API-Key": process.env.AIVA_API_KEY,
    "Content-Type": "application/json"
  },
  body: JSON.stringify({
    clientId: process.env.AIVA_MERCHANT_ID,
    message: "Where's my order?",
    channel: "WebChat",
    customerEmail: "customer@example.co.nz"
  })
});
const { message, confidence, requiresEscalation } = await res.json();

Full endpoint reference, channels (WebChat, WhatsApp, Email, SMS, Messenger, Instagram DM) and response shapes: API Documentation.

The MCP server

One config block connects your AI assistant to your store. Works with Claude Desktop, Claude Code, Cursor, Windsurf, Zed, Continue - anything that speaks MCP:

claude_desktop_config.jsonNPX · @GETAIVA/MCP
{
  "mcpServers": {
    "aiva": {
      "command": "npx",
      "args": ["-y", "@getaiva/mcp"],
      "env": {
        "AIVA_API_KEY": "YOUR_API_KEY",
        "AIVA_MERCHANT_ID": "YOUR_MERCHANT_ID",
        "AIVA_API_BASE_URL": "https://api.getaiva.ai"
      }
    }
  }
}
get_merchant_infoStore profile, plan and connection status
get_customersCustomer list with RFM segments attached
search_customersSearch customers by RFM segment
get_customerOne customer, in full detail
get_ordersOrders in a date range
get_subscriptionSubscription details
get_affiliateAffiliate details
analyze_customer_healthAI-powered customer health analysis
get_marketing_intelligenceObjective-based marketing insights
get_strategy_recommendationsBusiness strategy recommendations

Setup walkthroughs per tool: Claude Desktop · the full getting-started guide.

Build something real

Superprompts

Complete AI prompts that build production-ready integrations for you - paste one into Claude or Cursor and get a working app against the AIVA API:

NEXT.JS DASHBOARDREACT COMPONENTSSHOPIFY APPFLUTTER MOBILE APPN8N WORKFLOW
The Tech Build Blog documents how Aiva herself was built - 25 posts of build logs and engineering notes, straight from the workshop.
Browse the build blog