Microsoft’s vNext platform for D365 Contact Center is a major leap forward in creating a more robust, scalable, and AI-powered contact center experience. With a focus on integration with Microsoft Teams, AI-driven insights, and expanded communication channels, this platform offers significant advantages over its predecessor, Omnichannel for Dynamics 365 Customer Service. But what exactly has changed, and how does this impact your current setup?
Let’s break it down.
from omnichannel voice to d365 contact center: what’s new?
In July 2024, Microsoft introduced the fully-fledged D365 Contact Center (CCaaS), a new solution for managing customer interactions. This move follows Microsoft’s earlier investments, including the acquisition of Nuance and the introduction of the Digital Contact Center Platform in 2022. Fast forward to 2024, and the platform is much more than a voice extension for Dynamics—it’s now a comprehensive, AI-powered contact center solution.
I did my first implementation of the voice channel back in September 2022, during its early stages, and having done another implementation in 2024, I can confidently say the differences are night and day. Let’s look at some of the main differences.
key differences in the old vs the new infrastructure
1. Platform Integration and Core Technology
- Omnichannel Voice (Old Platform):
- Integrated as part of the Dynamics 365 Customer Service suite.
- Voice was one of several channels (email, chat, social), and the system was built around Azure Communication Services (ACS).
- Basic voice features, with Power Virtual Agents (PVA) used for IVR automation.
- Limited capabilities in advanced call center features like reporting, analytics, and agent performance management.
- D365 Contact Center vNext:
- Designed as a multi-modal platform for voice, chat, email, social media, and more.
- Features full integration with Teams, third-party telephony solutions (e.g., Genesys, Avaya), and other Microsoft technologies.
- Scalable, cloud-native architecture optimized for global deployments and minimal latency.
- Advanced AI-driven call center features like Automated Speech Recognition (ASR), real-time transcription, and sentiment analysis.
- Move from Power Virtual Agents to Copilot Studio bots for more powerful voice automation and handling complex workflows.
2. AI and Automation
- Omnichannel Voice relied on Power Virtual Agents for IVR and voice automation.
- The vNext platform integrates Copilot Studio bots, leveraging AI for real-time customer insights, dynamic voice routing, and suggestions during calls. It also enables advanced reporting and supervisor controls for managing agent performance.
While the Omnichannel voice platform leveraged Azure Communication Services (ACS) for cloud communication, it was built on a more monolithic architecture, which tied it closely to the Dynamics 365 Customer Service model. The vNext platform has transitioned to a cloud-native, microservices-based architecture. This enables much greater scalability, flexibility, and performance. Seamless cloud integration allows the platform to easily integrate with other Microsoft technologies (e.g., Teams, Azure Cognitive Services) and external telephony systems (e.g., Genesys, Avaya).
So in short; where you previously had limited voice architecture options, you now have a broader set of voice channel options. But if you, like my customers, are already onboarded with Azure Communications Services, you’d still need to make some adjustments to migrate your workstreams and channels onto the new infrastructure.
Existing workstreams will remain unaffected and will continue to function with already connected Power Virtual Agents (classic IVR bots), and they will be marked as type “Voice classic” in Customer Service admin center. Newly created workstreams will instead only be compatible with Copilot Studio Agents. Newly created workstreams will also need to be tied to new event grids for incoming calls and recording.
But, since the upgraded voice channels are only compatible with the Copilot Agents you need to create your new agents first!
step 1: create copilot studio agents to replace your power virtual agents
To help you upgrade your PVA to a Copilot Studio Agent there is a clone feature available which will clone the classic bot and create a new Copilot that uses the unified authoring version. It does, however, not include any topics that you might have built using Bot Framework Composer and any Bot Framework Skills that might be in use, will need to be reconnected with the new Copilot agent. When I cloned my PVA’s the custom topics did not transfer. Microsoft’s documentation state that custom entities should be transferred but I lost both them, and the topics in which they were used. So keep in mind that you might need to set some time aside to replicate your PVA dialogues in the new authoring canvas.
Once the clone is completed you need to re-configure the authorization model and the engagement channels to be used. If you used your PVA with dial tone input (DTMF) only it might be worth looking into setting up a variable in your agent to allow for DTMF only before you explore the speech functionality and experiment with the speech sensitivity.
step 2: test the copilot studio agents in a non-production voice channel
Before you assign the new Copilot Studio Agents to a live production workstream, it’s advised that you test them on a newly created workstream with a purchased Azure number. This way you can make sure that the Copilot Agent works as expected in the new infrastructure before you deploy them to production and assign them to a workstream that’s being upgraded.
step 3: upgrade the voice channels
To upgrade your existing workstreams and channels; update the event subscription in Azure for incoming calls and the recording webhook as per this guide. Copy the value in the Recording Web Hook Endpoint field in the Customer Service admin center –> Phone Numbers –> Advanced–> Manage Azure Communication Services (Preview) and paste it in the Endpoint field.
Now you can run the voice update. Navigate to Customer Service admin center and select Channels in Customer support. Select Manage for Voice update. On the Voice update page, all the workstreams with the associated phone numbers, inbound and outbound profiles, and the corresponding migration update and test status are displayed.
Select the required workstream and then select Run update. During the run, you’ll be given the option to replace or remove bots – in this step you can select the voice-enabled Copilot Agent that you created in step 1.
The time of the synchronization depends on whether you use Azure PSTN calling (direct offer) or direct routing, but give it at least 30 minutes, and then you’ll see the status of the finalized update on the Results page. You can then use the test status field to keep track of testing. If any call to any phone number attached to a profile or a workstream fails after 30 minutes, the application sets the Test status to Failed.
One last thing…
If your bots are assigned on queue, and not workstream level, and you have multiple ones to migrate – it might be a good idea to create a new capacity profile for inbound calls which you assign to all human agents and the new Copilot Studio Agents. This will allow you to keep your PVAs in the queue until all of them have been cloned, but by mapping the voice workstreams to the new capacity profile you’ll minimize the risk of your calls being handled (and dropped) by the old agents.
The D365 Contact Center vNext platform is a significant upgrade over its predecessor, offering greater flexibility, advanced AI capabilities, and integrated voice channels. Whether you are migrating from the old Omnichannel Voice or just starting, the new platform enables smoother, more intelligent customer interactions, with greater scalability for global operations.
The transition will require some adjustments, particularly if you’re already using ACS for voice, but the benefits of the new, cloud-native architecture and AI-powered automation are worth the effort.
If you have any tips from your own upgrade or want to share some thoughts in this, please feel free to leave a comment.
Lämna ett svar