Set up your first project

First, create your virtual agent. Open the dashboard and click the “New bot” button at the top.

You will be suggested to start from scratch or copy an existing project (if you already have one). For now, just choose "Start new" and click "Next".

Next, choose your primary language (it mostly affects the natural language understanding and natural language generation features), set the bot name, and (optionally) upload an avatar. This avatar will be used when your visitors will be interacting with your chat widget. Click "Next" when ready.

Now you can upload your business knowledge assets (URL links, PDF or DOC files, or conversation logs from another platform). We'll use that data to train and fine-tune the large language model (LLM) which will be able to answer questions that relate to this data. For exampl, if you upload a link to your F.A.Q., it will be automated instantly. If you send a link to your storefront, the bot will be answering questions about your products, etc.

Read more about knowledge management in Activechat

That's all! Once your virtual agent is ready, you will be able to play around with its built-in smalltalk feature immediately. Clicking the "Get started" button will take you to the Intents manager (we'll talk about it later), and you can click the "Test your chatbot" button above to open the self-hosted chat widget in a new window.

Your virtual assistant is pretty basic now, and can do only basic small-talk (try phrases like "How are you?", "Are you a bot?", etc).

Now it's time to customize this agent so that it can bring real value to your customers. But first, let's see how to use the live chat feature.