Cassidy offers two levels of custom AI instructions: user-level Personalization for tailoring your individual Chat experience, and organization-level Global Instructions for setting rules that apply to every Agent and Workflow. This article covers how to configure both.
Setting up user-level Personalization
Personalization lets you tell the AI about yourself so it can provide more relevant, tailored responses in your Chat conversations.
Open a new Chat: Click "+ New Chat" in the left sidebar or on the "Home" page. For more information about Chat, see Overview of Chat.
Open the Personalization modal: Click the "Personalize" button in the top-right corner of the new Chat screen.
Set your name or nickname: The "Name / Nickname" field is prefilled from your profile name. You can change it to whatever you'd like the AI to call you in Chat responses.
Enter your role: In the "Role" field, enter your job title or function (e.g., "VP of Sales" or "Software Engineer"). This helps the AI understand your context and tailor responses accordingly.
Add custom instructions: Use the "Custom Instructions" field to tell the AI about your preferences, work focus, or formatting expectations. For example: "I manage the enterprise sales team focused on AmLaw 200 firms. Keep responses concise and action-oriented. Use bullet points when listing key takeaways." This field supports up to 2,000 characters.
Save your settings: Click "Save" to apply your changes, or "Cancel" to discard them. A confirmation toast will appear at the bottom of the screen.

β
*Personalization settings only apply to your personal Chat conversations within the Cassidy app. They will not be included in Workflow runs or external deployments (Slack, Teams, Embed, API, etc.).
*Changing your name or nickname in Personalization does not change your profile name elsewhere on the platform. The Chat greeting will continue to use your profile name.
Setting up organization-level Global Instructions
Global Instructions allow admins to define AI behavior rules that apply to every Agent and Workflow across the organization. Use this to enforce your brand voice, set compliance guardrails, or provide important company context.
Open Organization Settings: Click your signed-in account name at the bottom of the left sidebar. Then, click the gear icon to the right of your "Current Organization" name.β
Navigate to Global Instructions: In "Organization Settings", click "Global Instructions" under the "Agents" section in the sidebar.
Add your first instruction: Click the "+ Add First Instruction" button. A new instruction card will appear with the following fields and controls:
A title field for labeling the instruction (e.g., "Banned Words" or "Company Description").
A text body for the instruction content (e.g., "Never use the terms 'guarantee,' 'legal advice,' or 'liability-free' in any AI output.").
A toggle to enable or disable the instruction without deleting it. Disabled instructions are excluded from the active character count.
A drag handle (three-line icon) on the left for reordering instructions.
A reset button (circular arrow icon) to revert the instruction to its last saved state.
A delete button (trash icon) to permanently remove the instruction. A confirmation dialog will appear before deletion
An "Apply to workflow actions" checkbox β unchecked by default.
Configure the "Apply to workflow actions" checkbox (optional): Check this box to apply the instruction to all "Generate" and "Generate with Agent" actions across your Workflows. Hover over the tooltip icon for details: "Applies this instruction to all Generate and Generate with Agent actions across your workflows. This will affect how existing workflows behave, so only enable this for universal instructions you want applied organization-wide."
Add more instructions: Click "+ Add Instruction" to create additional instruction cards. The total character counter at the top of the page tracks usage across all active instructions, up to a maximum of 15,000 characters.
Save your changes: Click "Save" in the top-right corner to apply all changes. A confirmation toast will appear at the bottom of the screen.
*Only Admins can access and configure Global Instructions.
*Global Instructions apply to every Agent by default, but are not included in Workflow actions unless you explicitly opt in per instruction via the "Apply to workflow actions" checkbox. This is off by default because it could significantly influence behavior in existing automations.
Examples of Global Instructions
Here are some examples of how you might use Global Instructions to standardize AI behavior across your organization:
Banned words/phrases: "Never use the terms 'guarantee,' 'legal advice,' or 'liability-free' in any AI output."
Company description: "SprinterSoftwareGroup is a leading provider of cutting-edge software solutions for legal professionals, specializing in case management, document automation, and client communication."
Formatting style: "Always use numbered lists for procedural steps and bullet points for non-sequential items. Use plain language and avoid legal jargon unless specifically asked."
Tone: "Maintain a professional, supportive tone. Address users as legal professionals and assume familiarity with standard legal workflows."
Compliance: "Do not generate content that could be construed as legal counsel. Always include a disclaimer when producing legal-adjacent content."
For more on writing effective AI instructions, see Editing an Agent and the Prompt engineering guide.







