How do I set up custom actions in Legal Server?

Custom actions let your AI agent perform tasks directly inside the conversation rather than just answering questions. Think of them as the bridge between chat and your legal software.

• Case creation: A site visitor asks about filing a new matter and the agent can capture intake details, then create a draft case in your Legal Server automatically.
• Status checks: Clients request an update on their matter. The agent can query your system and return the current stage - no human digging required.
• Scheduling: Book consultations by pushing appointment requests from the chat straight into your calendar tool.
• Government legal workflows: For public-sector teams, custom actions can kick off internal processes like FOIA request logging or mandated notifications.

Because every action is grounded in your own knowledge base, you control exactly what information the agent collects and how it interacts with your back-end tools.

Follow these steps to build a new custom action for your legal environment.

1. Open the custom actions builder
   Navigate to the agent settings in Chatref and locate the Custom Actions tab.

2. Define the trigger
   Give the action a descriptive name, such as “New Case Intake” or “Check Matter Status.” Choose what the user must say or ask to invoke the action (e.g., “I want to start a new case”).

3. Set up the data collection form
   Design the fields the agent needs to gather from the user. For case management customization, you might ask for client name, matter type, jurisdiction, and a brief description. Use drop-downs, text inputs, or date pickers - all configurable without code.

4. Connect your legal server endpoint
   Paste the webhook URL or API endpoint from your Legal Server into the action’s delivery settings. Chatref will send the collected data as a structured payload, which your system can then process. If you’re working with government systems, make sure the endpoint uses HTTPS and includes any required authentication headers.

5. Write the agent’s response
   Tell the agent what to say after the action runs. For example: “I’ve logged your new case and a lawyer will review it within two hours. Your reference is #12345.”

6. Test and refine
   Use the Chatref playground to simulate the action. The shared inbox lets a human team member watch the test in real time and intervene if anything needs adjusting.

Once saved, the action is live. The AI agent follows the same pattern on every matching request, hand-off to a human happens seamlessly if the user asks for something outside the defined action.

Government legal custom actions often involve compliance checks and multi-step approvals. With Chatref, you can layer in these controls.

• FOIA or public records requests: Create an action that gathers the requested document details and the requester’s contact info, then pushes the data into your tracking system as a new entry. An AI agent can handle the initial triage while the shared inbox alerts your records officer to any complex asks.
• Mandated notifications: Set up an action that automatically logs a notification event in your case management platform whenever a user asks about a notification deadline. The agent can confirm the log and inform the user, reducing manual entry.
• Internal escalations: Use custom actions to flag high-priority matters. If the agent detects certain keywords (e.g., “crisis,” “urgent,” “deadline”), the action can send a notification to a specific team inbox queue.

All government legal actions run against your knowledge base, so the agent remains consistent with your agency’s policies and tone.

Adopt these habits to make your case management customization stick.

• Keep actions focused: One action should do one clear job. A “New Case Intake” action should not also try to update a separate matter. Single-purpose actions are easier to maintain and less prone to error.
• Use field validation: Require the necessary information before the action proceeds. If a case needs a docket number, the agent should not push the data until it’s provided.
• Log everything in the shared inbox: The team can monitor every custom action execution from the Chatref inbox, giving you a full audit trail without extra effort.
• Review knowledge-base triggers: Your agent only knows what’s in your docs. Update the knowledge base regularly so the agent can correctly answer questions that lead into a custom action.
• Test with real scenarios: Run the action with sample data that mimics government or law-firm user flows. Pay attention to how the agent handles partial or unclear input - refine the action’s prompts to keep it on track.

What custom actions can I create in Legal Server?

You can create actions for case intake, status inquiries, document requests, scheduling, and compliance verifications. Any task that can be triggered by an API call - like creating a matter, updating a record, or sending a notification - can become a custom action in Chatref. For government teams, actions often cover FOIA logging, notification mandates, and inter-department escalations.

How do I automate tasks with custom actions?

Define the action’s trigger phrase and data fields in Chatref’s builder, then link it to your legal software’s endpoint. When a user says the trigger, the AI agent collects the needed information and pushes it to your system automatically. The shared inbox gives your team live visibility, and you can always hand off a conversation if the automation needs a human override.

What are the best practices for setting up custom actions?

Keep each action focused on a single task, enforce field validation so no incomplete data reaches your system, and test thoroughly with realistic user inputs. Monitor every execution from the Chatref inbox to catch edge cases, and update your knowledge base so the agent’s guidance stays accurate. Regular reviews help you adjust actions as your case management processes evolve.

How do I track custom actions in my legal case management system?

Every custom action is logged in your legal server via the webhook, creating a timestamped record. In Chatref, all conversations - including those that trigger an action - appear in the shared inbox with full context. You can also add a custom ID or reference number to the action’s response, making it easy to cross-reference chat interactions with the resulting case entries.