Add AI support to your website

Embed a chat widget that answers questions from your own content, no support team required, in about five minutes.

about 2 months ago
2 min read
374 words

Most websites answer the same handful of questions over and over — pricing, setup steps, refund policy, "does this work with X" — usually buried in a docs page, a FAQ nobody reads, or a support inbox. Basegent turns that content into an AI agent that answers those questions directly on your site, instantly, day or night.

Here's how to add it to yours.

1. Train it on your content

Before the widget can answer anything, Basegent needs something to answer from. In the dashboard, go to Ingest and point it at what you already have:

  • A documentation site or help center URL
  • A pricing page
  • FAQs, policies, or onboarding guides — pasted directly as text

Basegent fetches the content, compiles it into a searchable knowledge base (the Wiki), and keeps it around as the single source of truth your AI cites when it answers.

2. Test it before anyone else sees it

Use the Query playground to ask the exact questions your customers usually ask. Every answer comes with citations back to the wiki entries it used — if something's wrong or missing, fix the source entry and ask again. This loop is the fastest way to catch gaps before they show up as a bad customer experience.

3. Drop the widget on your site

Once you're happy with the answers, grab your embed snippet from Integration:

<script>
  window.Basegent = {
    tenantId: "<YOUR_TENANT_ID>",
    apiBase: "https://basegent.aevr.online",
    persona: { name: "Aria", greeting: "Hi! How can I help?" },
  };
</script>
<script src="https://basegent.aevr.online/widget/widget.iife.js"></script>

Paste it before the closing </body> tag and you're done. The widget renders in a Shadow DOM, so it won't fight with your site's existing CSS, and it's under 20kB gzipped — it won't be the thing that slows your site down.

4. Know when it doesn't know

Basegent doesn't guess. When a question falls outside your knowledge base, the AI says so and — if you've configured escalation — hands the conversation to a human via the live Inbox instead of making something up. You stay in control of what gets said on your behalf, and every unanswered question becomes a signal for what to add to your knowledge base next.

That's the whole loop: content in, answers out, gaps surfaced automatically. See the Getting Started guide for the full walkthrough, or Embed the Chat Widget for every configuration option.