Build a Chatbot Without Code: Complete Guide
You do not need to write a single line of code to build, train, and deploy an AI chatbot on your website. That was not true three years ago. In 2026, no-code chatbot builders let you go from zero to a working chatbot in under 30 minutes.
This guide walks you through the entire process: signing up, training your chatbot on your content, customizing its appearance and behavior, testing it, and embedding it on your site. Everything happens through a visual interface.
What No-Code Chatbot Builders Do in 2026
The current generation of no-code chatbot tools is fundamentally different from the decision-tree builders of a few years ago. Those older tools required you to map out every possible conversation path manually. Miss a branch and the user hit a dead end.
Modern no-code builders use large language models (GPT, Claude, Gemini) under the hood. Instead of scripting conversation flows, you train the chatbot on your content, give it instructions about tone and behavior, and the AI handles the actual conversation. It understands context, handles follow-up questions, and responds in natural language.
Here is what you can do without writing code:
| Capability | How It Works |
|---|---|
| Train on your website | Provide a URL; the tool crawls and indexes your pages |
| Train on documents | Upload PDFs, Word docs, or text files |
| Train on Q&A pairs | Type specific questions and their ideal answers |
| Choose an AI model | Pick from models like GPT-4o, Claude, or Gemini |
| Customize appearance | Set colors, position, welcome message, avatar |
| Set behavior rules | Write plain-language instructions for tone, scope, and escalation |
| Add actions | Configure lead capture, suggested messages, or custom buttons |
| Embed on any website | Copy and paste a single script tag |
| Support multiple languages | Chatbot auto-detects and responds in the visitor's language |
The things that used to require a developer, training an AI model, building a chat interface, handling real-time messaging, deploying to production, are all handled by the platform.
Seven Steps to Build Your Chatbot
Step 1: Sign Up and Create a Chatbot
Start by creating an account. On Agentkit, this takes about 30 seconds. You will land on a dashboard where you can create your first chatbot.
Click "Create Chatbot" and give it a name. Choose something descriptive, like "Acme Support Bot" or "Sales Assistant." The name appears in your dashboard for organization purposes; visitors on your website see whatever welcome message you configure later.
Select your AI model. Agentkit gives you access to a range of models across three providers:
| Provider | Models Available | Best For |
|---|---|---|
| OpenAI | GPT-4o, GPT-4o-mini, GPT-4.1, GPT-4.1-mini, GPT-4.1-nano, o4-mini | General purpose, strong reasoning |
| Anthropic | Claude Sonnet 4, Claude Haiku 3.5 | Nuanced conversation, detailed answers |
| Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.0 Flash, Gemini 2.0 Flash-Lite | Fast responses, multilingual strength |
If you are unsure, start with GPT-4o-mini or Claude Haiku 3.5. They are fast, capable, and cost-effective. You can switch models later without retraining.
Step 2: Add Your Training Sources
This is the most important step. Your chatbot's quality depends directly on the quality and completeness of its training data.
You have four source types to work with:
Website URLs: Enter your website URL and Agentkit crawls the pages. This is the fastest way to give your chatbot comprehensive knowledge. Point it at your marketing site, help center, documentation, or FAQ page. The crawler follows links on your domain and indexes the content it finds.
Tips for website training:
- Start with your most important pages: homepage, pricing, FAQ, product features
- If you have a help center or knowledge base, crawl that separately for thorough coverage
- Review the indexed pages to make sure nothing irrelevant was included
Documents: Upload PDFs, Word documents, or text files. This is useful for content that is not on your website: internal procedures, product specifications, training manuals, or policy documents.
Q&A pairs: Manually type a question and its ideal answer. This gives you the most control over specific responses. Use Q&A pairs for:
- Your top 10-20 most frequently asked questions
- Questions where the answer needs to be exact (pricing, policies, legal)
- Questions that your website content does not answer clearly
Plain text: Paste in any text content. Product descriptions, company background, team bios, or any other information you want the chatbot to know.
A good starting configuration:
| Source Type | What to Add | Priority |
|---|---|---|
| Website | Your main marketing site | High |
| Website | Help center or docs site | High |
| Q&A pairs | Top 20 FAQs with precise answers | High |
| Document | Product spec sheet or feature list | Medium |
| Plain text | Company overview, mission, key differentiators | Medium |
Step 3: Customize the Appearance
Make the chatbot look like it belongs on your site. In the customization panel, you can adjust:
Colors: Set the primary color to match your brand. The chat bubble, header, and accent elements all use this color. Enter a hex code or use the color picker.
Chat bubble position: Choose bottom-right or bottom-left. Bottom-right is the convention most visitors expect. Only change it if you have another element (like a support widget) already occupying that spot.
Welcome message: This is the first thing visitors see when the chat opens. Make it specific and helpful:
- Weak: "Hi! How can I help you?"
- Better: "Hi! I can answer questions about our product, pricing, and features. What would you like to know?"
- Best: "Welcome to Acme. I can help with pricing, getting started, or troubleshooting. What brings you here today?"
Bot name and avatar: Give the chatbot a name that fits your brand. "Acme Assistant" or "Support Bot" are clear and honest. Upload a custom avatar or logo to make it visually consistent with your site.
Initial messages and suggested prompts: Add three to five suggested messages that appear when the chat opens. These give visitors a starting point and guide conversations toward topics your chatbot handles well.
Step 4: Write Instructions
Instructions tell the chatbot how to behave. This is different from training data (what the chatbot knows). Instructions define the chatbot's personality, scope, and rules.
Write instructions in plain language. The chatbot follows them as behavioral guidelines. Here is a template:
Identity: "You are an AI assistant for [Company Name]. You help visitors learn about our product and answer their questions."
Scope: "Only answer questions based on the training data provided. Do not answer questions about topics unrelated to [Company] or its products."
Tone: "Be friendly, clear, and concise. Use simple language. Avoid jargon."
Limitations: "If you do not know the answer, say so honestly. Do not make up information."
Escalation: "If a visitor asks to speak with a human, wants to discuss billing issues, or has a question you cannot answer, let them know they can contact our support team at [email protected]."
Language: "Respond in the same language the visitor uses."
You do not need to be a prompt engineer. Write the instructions like you would write onboarding notes for a new team member. Clear, specific, and practical.
Step 5: Add Actions
Actions let your chatbot do things beyond answering questions. Agentkit supports five action types, and none of them require code:
| Action Type | What It Does | Example Use |
|---|---|---|
| Lead capture | Collects name, email, and custom fields | "Want me to send you a detailed comparison? What is your email?" |
| Custom form | Presents a structured form during the conversation | Application intake, support ticket creation |
| Custom button | Displays a clickable button in the chat | "Book a demo" button linking to your calendar |
| Custom API | Calls an external API endpoint | Check order status, look up account info |
| Suggested messages | Shows clickable prompts for the visitor | Guide visitors toward common topics |
For most chatbots, start with suggested messages and lead capture. These two actions cover the most common needs and take minutes to configure.
Step 6: Test in the Playground
Before putting your chatbot on your live website, test it thoroughly. Agentkit includes a playground where you can have full conversations with your chatbot and see how it responds.
Test these scenarios:
| Test Scenario | What to Check |
|---|---|
| Common question | Does it answer accurately from training data? |
| Question not in training data | Does it say it does not know, or does it make something up? |
| Multi-turn conversation | Does it maintain context across multiple messages? |
| Off-topic question | Does it stay within scope per your instructions? |
| Different languages | Does it respond in the visitor's language? |
| Lead capture trigger | Does the lead capture form appear at the right moment? |
| Edge cases | Rude messages, empty messages, very long questions |
Adjust your training data and instructions based on what you find. This iterative testing is the difference between a chatbot that impresses visitors and one that frustrates them.
Step 7: Embed on Your Website
Once you are satisfied with how the chatbot performs, it is time to go live. Agentkit gives you a single script tag to add to your website:
<script data-chatbot="your-chatbot-id" src="https://cdn.agentkit.ai/widget.js"></script>
Paste this before the closing </body> tag on any page where you want the chatbot to appear. That is it. No build step, no dependencies, no configuration files.
What You Can Customize Without Code
Here is a complete list of everything you can change through the Agentkit dashboard:
| Setting | Options |
|---|---|
| AI model | Models from OpenAI, Anthropic, and Google |
| Training sources | Website, document, Q&A, text |
| Primary color | Any hex color |
| Chat position | Bottom-right, bottom-left |
| Welcome message | Custom text |
| Bot name | Custom text |
| Bot avatar | Upload image |
| Suggested messages | Up to 6 clickable prompts |
| System instructions | Plain-language behavior rules |
| Actions | Lead capture, forms, buttons, API calls, suggested messages |
| Language | Auto-detect from 95+ languages |
| Domain restrictions | Limit where the widget loads |
| Rate limiting | Control messages per user per time period |
| Auto-retrain | Automatically re-crawl sources on a schedule (Standard+) |
All of these settings are available through point-and-click interfaces. Change them anytime; updates are reflected immediately without re-embedding the widget code.
When You Need the API
The no-code interface covers most use cases, but there are situations where the REST API (available on Hobby plans and above) gives you additional flexibility:
- Custom integrations: Connecting the chatbot to internal tools, databases, or workflows that Zapier does not cover
- Programmatic management: Creating or updating chatbots in bulk, managing training sources via scripts
- Custom analytics: Pulling conversation data into your own BI tool or data warehouse
- White-label deployments: Building the chatbot into your own product for your customers
- Advanced webhooks: Processing chatbot events with custom business logic
The API is not a requirement to get value from a chatbot. Think of it as an extension point for teams that outgrow the dashboard.
Platform-Specific Embedding
WordPress
Add the script tag to your WordPress site using a plugin like Insert Headers and Footers, or paste it directly into your theme's footer.php file. If you use a page builder like Elementor or Divi, add an HTML widget with the script tag. The chatbot works with any WordPress theme and does not conflict with other plugins.
Shopify
In your Shopify admin, go to Online Store, then Themes, then Edit Code. Open the theme.liquid file and paste the script tag before the closing </body> tag. Save and the chatbot appears on every page of your store. For specific pages only, use Shopify's section or page template system.
Squarespace
Go to Settings, then Advanced, then Code Injection. Paste the script tag in the Footer section. Squarespace loads it on every page automatically. The chatbot widget sits on top of your Squarespace layout without interfering with your design.
Maintaining Your Chatbot
Building the chatbot is step one. Keeping it useful requires ongoing attention:
Weekly: Skim conversation logs. Look for questions the chatbot could not answer and add that content to your training sources.
Monthly: Review the overall chatbot metrics. Are visitors engaging? Is the chatbot answering accurately? Update training data if your product, pricing, or policies have changed.
Quarterly: Do a full review of your instructions, suggested messages, and actions. As your business evolves, your chatbot should evolve with it.
Agentkit's auto-retrain feature (Standard plan and above) automatically re-crawls your website sources on a schedule, so your chatbot stays up to date as you publish new content. For Q&A pairs and documents, you will still need to update manually when information changes.
Getting Started
You can have a working chatbot on your website today. Sign up, add your website URL as a training source, write a few lines of instructions, and embed the script tag. The entire process takes less than 30 minutes.
Start with the free plan, which includes 50 messages per month and one chatbot, enough to test the experience and see how visitors interact. When you are ready for more volume, the Hobby plan at $29.99/month gives you 2,000 messages and API access. See our chatbot cost and pricing guide for how the plans scale as your usage grows.
No coding skills needed. No technical setup. No waiting for a developer.
No credit card required.
