Cassidy allows you to edit and fine-tune your AI Agents to ensure they align with your organization's specific requirements and brand voice. Follow these steps to modify your Agent:
Navigate to the Agents page: Click on the "Agents" option in the left sidebar of your Cassidy dashboard.
Open the Agent settings: Locate the Agent you want to edit and click the "..." icon in the top right corner of the Agent's card. From the dropdown menu, select "Edit."
Alternatively, you can edit an Agent directly from the chat interface by clicking on the Agent dropdown and selecting the "Edit" option next to the Agent's name.
Navigate between tabs: The Agent settings are organized into three main tabs:
Setup: Configure basic settings, AI model, instructions, Knowledge Base search, tools, and brand voice
Deployments: Deploy your Agent to external platforms
History: View chat history to monitor usage and see feedback from end users
Modify "Setup" tab configurations: Under the "Setup" tab, you can customize various aspects of your Agent:
Update appearance: Modify your Agent's icon, name and description.
Choose AI model: Select the base model for your Agent from available options.
Configure Advanced Model Settings: Click to expand the "Advanced Model Settings" section to fine-tune your Agent's behavior.
Max Mode: When enabled, the Agent will use the maximum possible input tokens for each chat message, allowing it to consider more of the conversation history and attached files. This may use more credits.
Temperature: Click to expand the "Advanced Model Settings" section to configure the temperature setting. Temperature is the parameter that controls the randomness of the results generated by your Agent. To change the temperature setting, click and drag the slider to adjust the value, or type in the text box to set an exact value. For non-creative tasks where you want the Agent to strictly adhere to instructions, it is best to use a lower temperature. You should use a higher temperature for highly creative tasks, such as marketing ideas, where there is room for the Agent to output dynamic results.
Adjust instructions: Add and remove instructions provided to the Agent to better guide its behavior and outputs. Clearly describe the tasks you want the Agent to perform and any specific guidelines it should follow. You can reference Knowledge Base items directly in instructions by typing the "#" key to search for and select items from your Knowledge Base.
When you reference an item using "#", the Agent will autonomously search over that content when relevant.
If you hover over a referenced item and click the pin icon, it will include the content directly in the prompt (counting towards token limits) - this is only possible if the content can fit inline. This means the Agent will always see the content in full.
Configure Knowledge Base search: Enable or disable the Knowledge Base Search feature, which allows the Agent to search the knowledge base. If enabled, select the specific parts of the knowledge base you want the Agent to access and choose between two search modes:
Always Search: The Agent will search the knowledge base every time, regardless of the request
Smart Search: The Agent will decide when to search the knowledge base. It only searches when it determines it's useful for answering the request
Limit search results timeframe: Enable or disable the Timeframe Limit for Search results, which allows you to select a specific timeframe for the search results within the knowledge base and prioritize those that are more recent and relevant.
When enabled, you will have the option to specify the timeframe on the chat page. Refer to the limiting search results timeframe article for more information.
Configure tools: Click "Add" to access the tools dropdown and select from available tools to enhance your Agent's capabilities:
Brand voice: After configuring tools, click the "Analyze" button to fine-tune your Agent's brand voice by uploading examples of your organization's writing style.
Preview your Agent: Optionally click the preview icon in the top right corner to chat with the Agent and test it out while iterating on your changes.
Set up external deployments: Click the "Deployments" tab to deploy your Agent to various platforms:
Save or publish changes: After making the necessary adjustments, save your changes or publish a new version of your Agent. For detailed information on publishing and managing versions, see Publishing Agents and reviewing version history.
Chat with Agent: Once your changes are saved or published, you can chat with your Agent to see your modifications in action.
By regularly editing and fine-tuning your AI Agents, you can ensure they remain up-to-date with your organization's evolving needs and consistently provide accurate, on-brand responses.