AI Assistants: Automating Routine Work

AI Assistants help automate routine management activities across your system. Instead of manually reviewing every case update, expense entry, or case review, you can create an AI Assistant that evaluates this content against criteria you define.

This saves managers time and helps enforce consistency across your team.

What Can AI Assistants Do?

AI Assistants are currently available for the following functions:

See below for detailed descriptions of each assistant type.

AI Assistant Types

Approve Expense Entry Assistant

Description

This AI Assistant reviews submitted expense entries and checks them against the criteria you define. 

Potential Use Cases

• Ensure required items (such as Mileage) don't exceed certain limits (i.e. greater than 50 miles).
• Enforce daily and weekly hour limits across all expense entries for an employee.
• Prevent excessive or unusual rates from being submitted unnoticed.
• Ensure certain items are submitted together (i.e. if mileage is submitted, travel time must also be included).

When it runs

The assistant evaluates submitted expense entries that match its Execution Rules (or all expense entries if no Execution Rules are set).

What happens when the Criteria is met

• The expense entry is automatically marked Approved.
• An internal note is added indicating it was approved by the AI Assistant.
• No further manager action is required unless you choose to manually review or make changes.

What happens when the Criteria is NOT met

• The expense entry is marked as Needs Attention.
• An internal note is added with the reason it could not be approved and suggestions to improve.
• A manager can use the dashboard or filter the expense entry list to see entries that are in the Needs Attention status.

Complete Case Review Assistant

Description

This AI Assistant reviews submitted case reviews and checks them against the criteria you define. 

Potential Use Cases

• Verify that all required questions have been answered appropriately
• Ensure field values are appropriate based on other responses (i.e. If the "Was video captured" was answered "Yes", then the "Minutes of video" must be greater than zero).
• Enforce standards for answer quality on important long-text questions
• Ensure reporting dates and key fields are filled in correctly
• Automatically approve fully completed questionnaires while sending incomplete ones to a manager

When it runs

The assistant evaluates submitted expense entries that match its Execution Rules (or all submitted case reviews if no Execution Rules are set).

What happens when the Criteria is met

• The case review is automatically marked Complete.
• An internal note is added indicating it was approved by the AI Assistant.
• No further manager action is required unless you choose to manually review or make changes.

What happens when the Criteria is NOT met

• The expense entry is marked as Needs Attention.
• An internal note is added with the reason it could not be approved and suggestions to improve.
• A manager can use the dashboard or filter the case review list to see entries that are in the Needs Attention status.

Validate Case Update Assistant

Description

This AI Assistant reviews submitted case updates and checks them against the criteria you define. 

Potential Use Cases

• Automatically approve routine surveillance or investigation updates that meet required formatting and content standards.
• Flag updates that are missing key details (such as video duration, location, or observations).
• Enforce consistent language and structure across field updates before they are shared with clients or managers.
• Catch incomplete or low-quality updates early so managers only review items that need attention.

When it runs

The assistant evaluates submitted case updates that match its Execution Rules (or all case updates if no Execution Rules are set).

What happens when the Criteria is met

• The case update is automatically marked Validated.
• An internal note is added indicating it was approved by the AI Assistant.
• No further manager action is required unless you choose to manually review or make changes.

What happens when the Criteria is NOT met

• The case update is marked as Needs Attention.
• An internal note is added with the reason it could not be approved and suggestions to improve.
• A manager can use the dashboard or filter the case update list to see case updates that are in the Needs Attention status.

How to Create an AI Assistant

1. Go to Settings > AI Assistants.
2. Click New AI Assistant.
3. Choose the type of assistant you wish to create.
4. Give it a clear name (example: “Surveillance Expense Entry Review Assistant”).
5. Optionally disable the assistant.  This is handy if you plan to add Execution Rules (see below) before activating the AI Assistant.
6. Select the Priority for this assistant. The system will execute the first assistant that matches all of the specified execution rules. If you have more than one assistant of the same type, prioritize targeted assistants with a smaller number (i.e. priority 1 will be evaluated before priority 2).  
7. Write your Criteria — these are the instructions the AI will follow.  Criteria should be written using markdown syntax here appropriate. 
8. Optionally, include the matching client policy (if applicable). 
9. Click Save AI Assistant.

Adding Execution Rules

By default your AI Assistant will run ALL items that require review. 

Execution Rules control which records an AI Assistant will process. If no Execution Rules are set, the assistant can run on all items of that type.  Add Execution Rules when you only want the assistant to review certain types of items.

1. Click Add Execution Rule to create a new run condition.
2. Click Add Condition and select the type of condition you wish to check for.  Different assistant types allow different conditions. 
3. Select the matching value that a record must meet to satisfy the condition.
4. Repeat this process until all conditions are entered.
5. Click Save Rule.

Once active, the assistant will automatically review matching records as they are submitted.

Best Practices

• Start with a small number of clear criteria and expand as needed.
• Test your assistant with a few sample submissions before relying on it fully.
• Use Execution Rules to avoid running the assistant on items that don’t need review.
• Review flagged items regularly so the system stays accurate.

Troubleshooting

• The AI Assistant is approving or rejecting items incorrectly.  Review the criteria defined in the assistant and try making it more specific and less vague. AI expects you to use the actual terms and/or item names in the system, so if its getting confused, be sure you're referring to the exact term or item you want it to evaluate.