Let's talk about low-code solutions. At CodeGPT, we have an integration with Make, where we can use the CodeGPT API to synchronize various services. For instance, let's assume we have documentation stored in Notion.

Our goal is to extract this documentation and insert it as knowledge base documents for an agent. I've created a Notion summary agent, and I'll start building a scenario. First, you can use any database where you have information or documents, and synchronize them with Make. Then, we invoke the integration with ChatGPT, which allows us to chat with the agent. We generate an exclusive API for this integration, paste it, and save the changes to establish the connection.

Once the connection is made, we identify the agent's ID and organization. We define the role, in this case, a user, and the content of the information that will be received as a response or message from the agent.

Finally, I want that summarized information to be sent to an email address. We integrate Gmail, and the information, which is the chat response from the agent, will be displayed in the email. After configuring all the parameters, we run the scenario and verify that the agent's output is correct according to the established parameters.

The information from Notion has been successfully received, and both the information received by the agent and the email content are accurate. After executing the scenario, we check the email to ensure the information is correct. We run the scenario once more, and indeed, we find the email received and sent correctly.