The HTTP Tool is a highly powerful feature, enabling AI Agents to execute HTTP Requests seamlessly, allowing you to call any API during a conversation. (Under the hood it uses function calling https://platform.openai.com/docs/guides/function-calling)

Xnapper-2024-01-12-10.30.52.png

This tool is integrated into the language model, which utilizes both the tool’s name and its detailed description, along with predefined parameters, to determine the appropriate moments for activation.

Additionally, the model can be specifically instructed on when to employ the tool through directives included in its system prompt.

For instance, an exemplary description of the HTTP Tool might be: “Useful for retrieving random cat images.”

Xnapper-2024-01-12-10.30.12.png

Request Parameters

The HTTP Tool editor facilitates the creation of necessary variables for conducting the HTTP request.

It includes an option labeled “Provided By User,” which signals to the model that it should gather the required data directly from the ongoing conversation.

Xnapper-2024-01-12-10.36.18.png

Streamlined Approval Process

With the activation of the “Approval Required” feature, the AI Agent will await authorization from an administrator before proceeding with the request.

This function ensures a controlled and secure operation.

Xnapper-2024-01-12-10.53.40.png

Xnapper-2024-01-12-11.17.42.png

Xnapper-2024-01-12-11.18.19.png

Xnapper-2024-01-12-11.19.14.png

Workflows

Moreover, the model is capable of executing multiple actions simultaneously and can integrate various tools to efficiently manage intricate workflows.