Configuration Panel

Overview

The Configuration Panel hosts various assistant configurations as well as the behavior of AI Studio itself. The most important is the ability to configure and review your assistant's Data Fields which include Assistant Session Fields and Configs.

You can switch between different configuration options by using the dropdown in the top left of the Configuration Panel


Data Fields

Assistant scoped fields that can be used within your AI assistant. A full overview on working with Assistant Fields can be found here.

Configuring Session Fields in the Configuration Panel

Configuring Session Fields in the Configuration Panel


Panels

The Panels section allows you to define both the default panel order, as well as whether or not your panels open in split view, or one at a time.

📘

The Conversation Panel is treated separately from other Panels, meaning it cannot be reordered, and it will always open alongside the panel you have open, regardless of panel behavior.

Configs

Configs allow you to set constant values that can be referenced throughout the assistant. Configs support several reference types such as Queue and Search Resource references as well as types similar to bot session fields (Text, Number, Boolean).


Configs can they be referenced throughout your Assistant for conditioning, message variables and within the python context off of the configs namespace. For example, configs.DEFAULT_ESCALATION_QUEUE in the rules editor or context.configs.BRAND_IDENTITY for python.

📘

Configs allow you to store and reuse content or logic throughout your experience while managing it from a single location. Common examples include privacy policy references, brand-specific language, or Queue settings for human agent handoff. When you update a Config, the changes automatically apply everywhere it's used.

Assistant Defaults

The default settings for your assistant, some of these, like the default LLM, can be overridden in specific scenarios


DefaultDescription
LLMSetting the default LLM allows all prompt configurations to inherit the same model. This can be overwriden on a prompt by prompt basis as required.