Get to your first reply in five minutes.

This guide walks you through creating a bot, training it on your knowledge, picking a model, and dropping the widget on your site.

1. Quickstart

Create an account at signup, name your first bot, then drop one URL or upload a PDF on the Sources tab. You can chat with the bot in the Playground immediately.

2. Adding sources

The Sources panel accepts websites (one URL or a sitemap), PDFs, .docx, .txt, .csv, Notion pages, and Google Drive folders. Each source can be re-indexed on demand or on a schedule. You can prune individual chunks if you want the bot to forget something.

3. Choosing a model

Open the Model tab and pick from Anthropic Claude, OpenAI GPT, Google Gemini, Groq-hosted Llama, or Mistral. You can override the model per bot and define a fallback chain for the unlikely case a provider has an outage.

4. Embedding the widget

Paste this snippet right before </body> on any page:

<script
  src="https://cdn.aisnschat.com/widget.js"
  data-bot-id="YOUR_BOT_ID"
  defer>
</script>

For server-side rendering or React, see the SDK section below.

5. REST API

All actions in the dashboard are also available via API. Authenticate with a bot key from Settings → API.

curl https://api.aisnschat.com/v1/chat \
  -H "Authorization: Bearer $AISNS_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "bot_id": "bot_8j2K9mNqLp",
    "message": "How do refunds work?",
    "session_id": "user_42"
  }'

6. Webhooks

Subscribe to message.created, handoff.requested, and lead.captured to wire Ai-Sns.io into your own systems. Webhooks are signed with HMAC-SHA256 so you can verify authenticity.

7. SDKs

Official packages for JavaScript/TypeScript, Python, Go, and Ruby. All four mirror the REST API one-for-one and include streaming helpers.

// npm install @aisnschat/sdk
import { Client } from "@aisnschat/sdk";
const client = new Client({ apiKey: process.env.AISNS_KEY });

const reply = await client.chat({
  botId: "bot_8j2K9mNqLp",
  message: "What's your return policy?",
});
console.log(reply.text);