Agent - Model Configuration
This documentation provides instructions for deploying and configuring Agent models.
An Agent model is a configuration of a Large Language Model (LLM) that outlines the specific features and functionalities to be included in the deployment and determines which capabilities of the LLM that are to be utilized.
The configuration required depends on the product:
Personal Agent – Comes with models maintained by Ayfie.
Business Agent SaaS – Comes with models maintained by Ayfie; however, a tenant admin can override them.
Business Agent On-Premises – Comes with the system models when installing and upgrading; however, an admin can add additional custom models following the instructions below.
Index – Comes with no configured models, and an admin must configure and maintain them following the instructions below.
Agent models come in two variants:
Agent System Models - these are predefined models. An admin only needs to provide the Azure API token and API address. Currently, all system models are Azure OpenAI LLMs hosted in the EU. See Agent - System Models Configuration for instructions on how to deploy the models in Azure and configure them.
Agent Custom Models - These models require manual configuration by an admin and can be set up for any supported LLM. See Agent - Custom Model Configuration for instructions on how to deploy and configure these models.
For both variants, the admin must maintain the models and keep them up to date. New Agent releases may include updated system models. See Agent - System Models Configuration for information on how to update them.