From invitation to a live chat

IperChat is a small private service. The chat for your website is set up by the administrator, the person who runs this IperChat service, and you are invited to manage it. There is no self-service registration. Three phases: get invited, customize, publish.

1. The administrator creates your chat

The administrator creates the chat for your website and invites you with your email address. Behind the scenes this creates your dedicated space, the tenant, with its own settings, documents and conversations. Nothing to do on your side yet.

2. Sign in to the operations panel

Go to admin.iperchat.ai:

1. Enter the email address you were invited with.
2. You receive a sign-in link by email (check spam if it doesn’t arrive within a minute). It works once and expires after 15 minutes.
3. Open it, on this device or on your phone: the panel page where you asked for it signs you in by itself. There is no password to choose: next time, ask for a new link.

3. Customize your chat

In the panel, open “Agent Configurations” (side menu → Agents group) and click the row of your chat, called an origin. This is where you set colors, name, greeting, suggested questions and the AI model: see Customize colors, greeting, and button.

To make the agent answer from your own material (price lists, FAQs, manuals), add it in the “Knowledge Base” section: see Upload documents, text, and images.

4. Put the chat on your website

In the “Documentation” section (side menu → Agents group) you’ll find the snippet ready to copy, with your Tenant ID already filled in:

<script
  src="https://widget.iperchat.ai/iperchat.iife.js"
  data-iperchat-auto-init
  data-tenant-id="YOUR_TENANT_ID"
  data-position="bottom-right"
></script>

Paste it into your site before the </body> tag. Reload the page: the chat button appears in the bottom right. For WordPress there’s a dedicated plugin: see WordPress plugin in 5 steps.

Limits

The administrator sets the limits of your chat: a monthly credit limit, and the knowledge-base limits (number of documents, storage, size per file). A credit is a unit of AI usage: with most models one reply costs one credit, see Choosing your LLM model. If you hit a limit, ask the administrator to raise it.