In this article, we’ll walk you through the first step: creating your AI Agent. You’ll set up the foundation, test it in the playground, and prepare it to go live.

The initial setup — including training your AI Agent on your company’s data — typically takes between 30 minutes and 3 hours. Most teams see a return on that time investment within just a few days of setting their AI Agent live.

Once your AI Agent is ready, you can add it to an automation to make it accessible to your customers (covered in a separate article). In Automations, you’ll decide when and where the AI Agent should respond — giving you full control.

1. In the navigation menu, click on AI Agent.

2. Click on New AI Agent.

3. In the setup modal, describe what type of AI Agent you want to create and what it should do — for example, “Create a support agent that helps customers with product questions.” The more detailed your description, the better the first version will be.

4. Optionally, add knowledge now by entering your website URL or uploading internal documents to train your AI Agent on your company’s data. You can also update or expand the knowledge later at any time.

5. Within seconds, Superchat will generate your fully configured AI Agent — ready to test in the playground.

Once your AI Agent is created, you can fine-tune it in the AI Agent settings to make sure it perfectly fits your company’s needs and brand. This is where you define how it behaves, what it knows, and what it can do before making it public to customers.

In this section, you define who your AI Agent is and how it communicates. It defines how your AI Agent behaves — not what it knows. Keep factual or product-specific information out of the prompt and put it in Knowledge & Data instead. Think of the prompt as your AI Agent’s job description and the knowledge as its training materials.

Name and Avatar

Choose how your AI Agent appears in Superchat (only visible to your team).

Prompt

The core setup that defines your AI Agent’s personality, tone, and behavior. It consists of three parts:

By default, every AI Agent already:

Here you provide the information your AI Agent relies on to answer customer questions accurately. Think of it as your AI Agent’s memory — everything it needs to know to respond accurately. The quality of your data directly impacts the quality of your answers — the better your data, the better your results.

You can add knowledge in three ways:

Connect your website to let your AI Agent learn from your content. You can add either:

After scraping, you can choose which pages should be accessible to your AI Agent.

To refresh your data, click the ⋯ icon and select Sync domain or Sync page — this updates your content with the latest version from your website.

Keep in mind:

Upload internal files to expand your AI Agent’s knowledge base. Supported file types: .pdf, .docx, .xlsx, .txt, .csv. Maximum file size per document: 25 MB.

Typical uploads include: Product catalogs, Terms & policies, Help articles, Training materials.

Best practices:

Create structured lists of question-answer pairs for common topics. Each QA list acts like a folder — for example:

QA lists are the most reliable way to provide data to your AI Agent. They’re easy to manage, structured, and instantly updatable inside Superchat — making them the foundation of a high-performing AI Agent. Over time, move your company data step by step into QA lists for the best long-term results.

Tools let your AI Agent do more than just chat — they enable actions inside Superchat. You decide which tools your AI Agent can use.

Available Tools

If your AI Agent should use tools, mention them in the prompt under Role & Tasks.

Examples:

You can test your AI Agent at any time using the Playground on the right side of the screen. When your AI Agent performs an action — like writing an internal note — you’ll see it appear directly in the playground. This helps you confirm everything works correctly before activating your AI Agent in an Automation.