Quickstart

This walks you from zero to a working /ask call.

Create a project

In the Console, create a Project (a tenant in the backend). This is the isolation unit that owns your model config, tools, knowledge, and conversations.

Configure a model

Open AI Settings and pick a model (e.g. gpt-4.1-mini), a temperature, and toggle Tools / RAG on if you need them. See Core concepts for what each does.

Grab your credentials

From the Console, copy your Tenant ID and generate an API key (egoz_live_… or egoz_test_…).

Make your first call

npm install @ego-z/client

import { EgoZ } from '@ego-z/client';

const egoz = new EgoZ({
  apiKey: process.env.EGOZ_API_KEY,     // egoz_live_xxx
  tenantId: process.env.EGOZ_TENANT_ID,
});

const res = await egoz.ask({ message: 'What is the refund policy?' });
console.log(res.answer);
console.log(`Tokens used: ${res.usage.totalTokens}`);

curl -X POST https://api.egoz.io/egoz/ask \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EGOZ_API_KEY" \
  -d '{
    "tenantId": "'"$EGOZ_TENANT_ID"'",
    "message": "What is the refund policy?"
  }'

Test & monitor

Use the Console Playground to iterate on prompts and tools before production, then watch usage, cost, and conversations in Analytics and Threads.

Response shape

Every /ask reply includes answer, threadId, the classified intent, toolUsed, ragChunksUsed, usage (token counts), and model. Full details in the Client SDK reference.

Next steps

Keep a conversation stateful → threading.
Stream tokens as they generate → streaming.
Let the AI call your APIs → tools in the Console.
Ground answers in your docs → knowledge base.