In this Use Case, we’re building a Copilot Agent to manage callback requests efficiently. The Agent automatically converts any callback request into a task. It supports outbound campaigns in Nimbus and can create or schedule a callback task whenever a customer asks to be contacted. Using natural language understanding, the Agent detects and extracts the preferred date and time from the customer's message.
Preconditions
To implement this Use Case you need:
-
For Nimbus:
- A Contact Center or Advanced Routing service license, which enables the use of advanced Workflow Activities.
- A Virtual Users license to integrate Nimbus with Copilot AI services.
-
For Microsoft:
- A “Copilot Studio” Licence → Also see Azure Billing notes below.
- Power Automate licensing needs, and all other Nimbus Power Automate Connector preconditions must be met.
- A deployed AI model or Power Automate Premium license to use AI Builder.
- Tenant Admin rights are required to implement all actions and connect to data sources required for your Services.
Overview of the Callback Agent
Designing the Copilot
Let's start by applying the best practices to our creation process by defining a scenario:
This bot is designed to transform any callback-related request into a manageable and trackable task within Nimbus. It is especially useful when your customers call from various locations and time zones, and you need a reliable way to handle their requests without losing oversight.
The bot uses advanced natural language understanding to interpret callback requests, detect and validate date and time information, and automatically create or schedule callback tasks. Whether the request is part of an inbound inquiry or an outbound campaign, the bot ensures every callback is captured and processed accurately.
Operating within the Virtual User (VU), the bot oversees all planned and scheduled tasks, allowing it to plan, initiate, or reschedule callbacks as needed. This ensures seamless coordination across different teams and time zones.
We this bot when you want to automate and control callback handling, eliminate manual tracking, and ensure no customer request falls through the cracks.
| Role |
The Callback Assistant is a Virtual User designed to manage and automate customer callback requests. Think of it as a reliable team member whose job is to ensure no callback request is missed, misunderstood, or delayed. Its responsibilities include:
|
|---|---|
| Tasks |
This Virtual User operates with limited autonomy: it can read and write to a specific SharePoint lists, initiate calls via Nimbus, and interact with customers—but only within the scope of its defined tasks. Each task is implemented as a tool in Copilot Studio, triggered by specific customer intents:
Each tool is designed with clear entry conditions and limited permissions, ensuring the bot only acts when appropriate and never oversteps its role. |
| Access, checkpoints and thinking |
The Callback Assistant only accesses:
It does not access CRM systems, customer history, or sensitive data unless explicitly required and approved. This ensures data minimization and responsible automation. To maintain control and oversight callback scheduling could include a confirmation step before the callback is initiated. This could be done by the tool feeding callbacks into Nimbus from the Sharepoint list. It would be sufficient to mark a callback entry as "created by Copilot" and then check on this column when feeding the callbacks in. We do not implement this, but you may consider it when implementing such an agent. 🔎All actions are logged for transparency and auditability using Azure insights. |
Building the Copilot Agent
| Description | Screenshot |
|---|---|
|
 |
|
Create a custom topic “Process JSON Input”. he goal is to hand over the customer's JSON to process You can follow the instructions in Luware Nimbus - Copilot - Initial Data Handover from Nimbus to Copilot to build the topic.
|
 |
| At the end of the “Process JSON Input” topic, redirect to a new topic called “Greeting”. |  |
|
Customize the custom topic “Greeting”.
💡The goal is to greet a customer by name, whenever the 🔎 This variable matches to System Fields > Call data parameter |
 |
|
 |
| ✅From this point on the base functionality is done. You can adjust all other topics according to your needs. | |
Build the tools
✅As this Copilot has “Orchestration” enabled, the logic is build into the tools. Let's go ahead and create the tools.
Start an immediate call back
Description
Start call back immediately. Only run this tool when there is urgency to immediately start a call back rather than scheduling one at a certain time. Use the global variable phone number as input or ask for user input if the variable has no value.
Find call back entries for a given phone number
Description
Use this tool to retrieve the OutboundTaskState value and respond to the question:
“When will you call me back?”
This tool checks the callback status using the OutboundTaskState field. Interpret values as follows:
Scheduled: Callback is planned — respond with “You have a callback scheduled. We’ll contact you soon.”
In Progress: Callback is ongoing — respond with “We’re currently trying to reach you.”
Destination Accepted: Callback was successful — respond with “We’ve already called you back successfully.”
Max Retries Reached, Removed, or Failed: Callback was unsuccessful — respond with “We tried calling but couldn’t reach you.”
If the state is unknown or missing, respond with “Callback status is unclear. Please check again later.”
Inputs
Select your Sharepoint list and set the filter query to custom with the following powerfx formula
"Number eq '" & Global.PhoneNumber & "'"Add a call back entry to the scheduled call back list
Description
Use this tool to schedule a new callback when the user says things like: “Call me back”, “Schedule a callback”, or “I didn’t get a callback.”
Parse natural language datetime (e.g. “in 2 hours”, “tomorrow at 6”) using the current time in the user’s timezone (UTC+2).
If the input is ambiguous (e.g. missing AM/PM or date), ask the user to clarify.
Store the callback time in UTC.
Do not use this tool for urgent or immediate callbacks.Inputs
Use the global variables as entry values. Adjust to your needs.
Finally, respond with a fixed message which will also act as the exit criteria in the Nimbus Virtual User workflow activity.
Adding the Copilot to Nimbus
Bot Authentication
✅The following steps explain how to define your authentication in Copilot Studio so Nimbus can access the API later.
- Log into Copilot Studio.
-
Security: Go to Agents > Settings > Security.
- Select No authentication → See Design and Limitation Notes below.
- Go to Web Channel Security > Secrets and Tokens.
- 🧠Copy either Secret 1 or 2 for later within your Nimbus Bots configuration.
- Ensure “Require secured access” is enabled.
Copilot API Endpoint
| General1 | https://directline.Botframework.com/v3/directline/conversations |
|---|---|
| Europe | https://europe.directline.botframework.com/v3/directline/conversations/ |
1 ☝A request might fail if you use the global base URI for a regional bot, as some requests could go beyond geographical boundaries. The URLs are maintained by Microsoft and retrieved from the Learn | Direct Line documentation. We advise to test performance and stability.
🧠Don't forget: Note down the endpoint for later use, e.g. for the Bots configuration of your Nimbus Virtual Users.
Configuring the Bot
✅The bot contains the API key and endpoint for your Copilot agent:
- Head to Nimbus Administration > Configuration > Bots and create a new Bot.
- Give your Bot a descriptive name.
💡This name is just for Nimbus UI purposes, and must not match Copilot. -
Define the Organization Unit under which this Bot will be selectable.
☝ Note that any service on this OU or higher can access it (which may result in additional Azure cost). - Type is M365 Copilot Direct Line 3.0.
- API Key: 🧠 Paste the Secret from the Copilot setup above.
- Endpoint: 🧠 Paste the Endpoint URL from the Copilot setup above.
Create a Response Template
- In Nimbus, go to Admin > Configuration > Bot Response Templates
- Add a new default response template for Copilot.
Create the Virtual User
- In Nimbus, go to Admin > Configuration > Virtual Users.
- Add a new Virtual User with a Description and choose your bot accordingly.
- On the Initial Message to Bot we handover the following data:
{
"Phone": "+$(Caller.TelNumber)",
"Name": "$(Customer.DisplayName)"
}
Using the Copilot in Nimbus
The final step is to add your newly defined Virtual User to a workflow.
Use the Virtual User in your Workflow
- Within Nimbus go to Configuration > Workflows.
- Create a new Audio/Video Workflow or add your new “Add Virtual User” activity to an existing workflow.
- Within the activity, you need to configure the Exit with the Regular Expression
^Goodbye$to ensure that the Virtual User stops after processing the second input.
💡The wording “Goodbye” is adjusted in Copilot.
- Within the activity, you need to configure the Exit with the Regular Expression
