How the workspace works
The screen is split into two main areas. On the left, you configure the behavior of the AI Agent. On the right, you test the agent in real time and immediately see how it responds.
Use natural language to train behavior
In the chat area, write what you expect the agent to do, what it should avoid, how it should respond, and what goal it should support. You can refine the behavior until it feels right.
Add knowledge sources
· Upload files such as PDFs, documents, books, course transcripts, or guides.
· Add links such as a website or Shopify store.
· Paste raw text directly into the system when you want to add or edit large blocks of knowledge manually.
Restrict the agent to your knowledge base
Because the agent is powered by advanced AI models, it may sometimes answer questions beyond the content you provided. If you want to prevent that, enable the restriction so the agent only responds based on your uploaded knowledge.
It is also a good idea to instruct the agent to clearly say when a question is not covered in the knowledge base.
Use templates
If you do not have content ready, you can use ready-made templates from CommuniPass. This can help you create an AI Agent quickly and then customize it for your audience or business.
Test and iterate
After each change, test the agent on the right side. Try common user questions, edge cases, and questions that should not be answered. Keep refining until the responses are accurate, useful, and aligned with your goal.
Related articles
· AI Agents: Overview
· How to Create Your First AI Agent
· AI Agent Quality Checklist
· How to Connect Your AI Agent to WhatsApp, Website, Messenger, and Instagram DM