Quick Start

Build your first AI assistant in Curatu and get it ready to answer questions. This guide takes you through the core setup: create a Knowledge Base → add content → test your assistant → deploy it on your website.

Once your Knowledge Base is ready, you can also connect your assistant to other channels or use a Workflow for more advanced experiences.

What You Need

Before you begin, make sure you have:

• A Curatu account with access to the Knowledge Base.
• Content for your assistant, such as PDFs, Word documents, or public help-center URLs.
• Access to Channels if you want to deploy your assistant on a website or another supported channel.
• The required plan and permissions for features such as Workflow, Chat Canvas, and Channels.

The features available in your workspace depend on your plan and role. If a feature does not appear in the sidebar, it may not be included with your current access.

Step 0 — Create Your Account

If you are creating a new Curatu organization, complete the signup process before continuing.

Create a New Organization

1. Open the Curatu signup flow and choose a plan.
2. Enter your Personal Details.
3. Complete OTP Verification using the code sent to your email or phone.
4. Add Payment details if required for your selected plan.
5. On You’re All Set, review your trial or plan details.
6. Sign in to your account.

After signing in, Dashboard is your home page. Use the sidebar to access Build, Deploy, and Monitor.

Join an Existing Team

If you received an invitation to join an existing organization:

1. Open the invitation link from your email.
2. On the Onboarding page, enter your first name, last name, job title, and password.
3. Select Register.
4. Sign in using your new credentials.

Step 1 — Create a Knowledge Base

A Knowledge Base stores the information your AI assistant uses to answer questions. You can add documents, URLs, and other supported sources to provide the assistant with information about your product, services, or organization.

1. Sign in to Curatu.
2. Open Build → Knowledge Base.
3. Select Create.
4. Enter the required Knowledge Base details.
5. Add a Name to identify the Knowledge Base.
6. Optionally add Instructions to define how the assistant should respond and what information it should focus on.
7. Select the Model and Temperature settings.
8. Save the Knowledge Base.

After it is created, you are taken to the Knowledge Base details page.

Tip: For your first Knowledge Base, start with a small and focused set of content. This makes it easier to test the assistant and identify any content that needs improvement.

Step 2 — Add Content

After creating your Knowledge Base, add the information your assistant should use when answering questions.

1. Open the Knowledge Base.
2. Select Add Data.
3. Choose the type of content you want to add.

For a first setup, Upload Files is a simple way to provide content quickly.

4. Add your files or URLs.
5. Close Add Data after your content has been added.
6. Wait for the content to finish processing before testing the assistant.

Important: Test your assistant after your content has finished processing. Testing before processing is complete may not reflect the information you have added.

Step 3 — Test Your Assistant

Use the Knowledge Base Test experience to check how your assistant responds before making it available to users.

1. Open your Knowledge Base.
2. Select Test.
3. Ask questions that your customers are likely to ask.
4. Review the answers.
5. Check whether the answers are relevant and based on your Knowledge Base content.

If the answers need improvement, consider adding more relevant content or updating the Knowledge Base Instructions in Settings.

Test again after making changes to confirm that the assistant responds as expected.

Step 4 — Deploy Your Assistant

Once your assistant is answering questions as expected, you can make it available to users through a supported channel.

For a website deployment, use the Web channel.

Create a Widget Design

Before creating the Web channel, create the appearance of your website chatbot.

1. Open Deploy → Channels.
2. Open the Web channel section.
3. Select Widget Design.
4. Create a new Widget Design.
5. Configure the appearance, branding, colors, messages, and other available options.
6. Save the Widget Design.

The Widget Design controls how the chatbot appears on your website. It is separate from the Knowledge Base that provides the chatbot’s information.

Create a Web Channel

After creating your Widget Design:

1. Open Deploy → Channels.
2. Create a Web channel.
3. Enter a name for the Web channel.
4. Select your Knowledge Base as the source.
5. Select the Widget Design you created.
6. Add the Allowed Domain where the chatbot will be used.
7. Create the Web channel.

Deploy the Web Channel

When you create a Web channel, it starts in Draft state.

The channel remains in Draft until its embed code is used on a website. Once the embed code is used and the chatbot is accessed through the website, the channel automatically moves to Deployed.

No manual deployment action is required.

Add the Chatbot to Your Website

1. Open the Web channel.
2. Select View Code.
3. Copy the provided embed code.
4. Add the code to your website.
5. Open your website and test the chatbot.

The Web channel must be configured with an allowed domain and a Widget Design before it can be used on your website.

Step 5 — Monitor Conversations

After your chatbot is available to users, use Monitor to understand how people are interacting with it.

Review Individual Conversations

Open Monitor → Conversations to browse individual conversation sessions.

You can search and filter conversations, review conversation details, and inspect information such as resolution, feedback, categories, and channel information.

Generate Conversation Reports

Open Monitor → Reports to generate written summaries of conversations for a selected date range.

You can use Date Range, Feature, and Category filters to define the conversations included in a report.

Optional — Use a Workflow

A Workflow lets you build more advanced conversational experiences using multiple connected nodes and logic.

If Workflow is available for your organization:

1. Open Build → Workflow.
2. Select Create.
3. Build your workflow in the visual editor.
4. Connect the required nodes.
5. Test and configure the workflow.
6. Use the workflow when configuring a supported channel.

Workflows can be used when a simple Knowledge Base experience is not enough for your use case.

Optional — Use Chat Canvas

Chat Canvas lets you create structured visual responses that can be displayed inside chatbot conversations.

To use Chat Canvas:

1. Open Build → Chat Canvas.
2. Create a Canvas.
3. Configure the Canvas content and appearance.
4. Save and deploy the Canvas.
5. Use the Canvas in a supported Workflow.

Chat Canvas is designed to work with a Workflow and an Agent Node.

Quick Reference

Troubleshooting