Adding an AI chatbot to your website used to mean months of development work, a team of ML engineers, and a budget that only enterprises could afford. That's changed. Today you can deploy a chatbot trained on your own documents in the time it takes to make a coffee.

This guide walks through the exact steps to go from zero to a live chatbot on your website — using NuovaBot.

What you'll need

  • Your support content in any of these formats: PDF, DOCX, TXT, Markdown, or CSV
  • Access to your website's HTML (to paste one script tag)
  • About 10 minutes

Step 1: Create your account and first chatbot

Sign up at nuovabot.com/sign-up. The free plan gives you one chatbot and 25 messages per month — no credit card required. When you sign in, you'll land on the chatbot dashboard. Click New Chatbot and give it a name (e.g. "Support Bot") and an optional description.

When you create the chatbot, NuovaBot generates a unique API key. Copy it — you'll need it for the embed snippet later.

Step 2: Upload your documents

Go to the Knowledge Base section. Click Upload Document and add your content. Good starting materials:

  • Your FAQ page exported as a PDF or text file
  • Product documentation or spec sheets
  • Your return or shipping policy
  • Onboarding guides or help center articles

NuovaBot processes each document automatically: it splits the content into searchable chunks, generates vector embeddings, and indexes everything so the chatbot can retrieve the right information when a user asks a question. Processing typically takes under 30 seconds per document.

You can upload as many documents as your plan allows and update them at any time. The chatbot's knowledge updates immediately after processing — no re-deployment needed.

Step 3: Configure the chatbot's behavior

Go to the Settings tab on your chatbot. The key fields:

  • Name — what the chatbot introduces itself as (e.g. "Aria from Support")
  • Persona — describe the tone and personality (e.g. "Friendly, concise, professional. Always offer to create a support ticket if you cannot answer.")
  • Welcome message — the first thing users see when they open the chat
  • Fallback message — what the chatbot says when it can't find a relevant answer in its documents
  • Knowledge mode — choose "Documents Only" to ensure every answer is grounded in your uploaded content, or "General Knowledge" to allow the chatbot to supplement with broader knowledge

A well-written persona and fallback message go a long way. Something like "I couldn't find a specific answer — let me create a support ticket and a human will follow up within 24 hours" converts what would be a dead end into a lead capture.

Step 4: Embed the widget on your website

Go to the Embed & Deploy tab. You'll see a script tag that looks like this:

<script
  src="https://nuovabot.com/widget.js"
  data-chatbot-id="your-chatbot-id"
  data-api-key="your-api-key"
  defer
></script>

Paste this tag into your website's HTML, just before the closing </body> tag. That's it. The floating chat button will appear in the bottom-right corner of every page.

The widget works on any website or CMS — Shopify, WordPress, Webflow, custom HTML, or anything else. It renders inside a Shadow DOM, which means it won't conflict with your existing styles.

Step 5: Test it

Before sending it live, use the Playground tab in your dashboard to test the chatbot with real questions. Ask things your customers typically ask and verify the answers are accurate and well-phrased.

If the chatbot misses an answer, the fix is usually to add more specific content to the knowledge base, not to reconfigure the AI. The quality of the answers is directly tied to the quality and completeness of your uploaded documents.

Optional: enable actions

On Starter plans and above, you can enable Actions — capabilities that let the chatbot do things beyond answering questions:

  • Ticket creation — the chatbot creates a support ticket and notifies your team by email when a user reports an issue
  • Meeting booking — the chatbot shares a Calendly link (or collects details and emails you) when someone asks for a demo
  • Live agent handoff — fires a webhook to Intercom, Crisp, or a custom endpoint when someone asks to speak to a human

What happens next

Once live, every conversation is logged in the Analytics section. You can see what users are asking, which questions the chatbot couldn't answer (a direct signal of what content to add next), and how much of your monthly message quota you're using.

The most impactful thing you can do post-launch is spend 30 minutes a week reviewing unanswered questions and adding the relevant content to your knowledge base. Each iteration makes the chatbot measurably better.

Ready to get started? Create your free chatbot — it takes under 10 minutes.