The Evolution of Power Virtual Agents to Copilot Studio

With this year’s Microsoft Ignite event we received many exciting announcements – one of which was Power Virtual Agents getting rebranded to Copilot Studio!

 

The Evolution of Power Virtual Agents

Before we delve into what has changed, it’s important to explain exactly what Power Virtual Agents is. Power Virtual Agents (PVA), a component of the expansive Power Platform, stands as a cloud-based software service proffered by Microsoft.

It allows users, even those without programming knowledge, to create and deploy sophisticated AI-powered virtual agents, commonly known as chatbots. Chatbots streamline customer service and internal workflows with their natural conversational abilities.

 

It is also important to touch on the history of PVA and what has led to this rebranding in the first place. In December 2019, as we unveiled PVA’s general availability, the surge in generative AI that followed was beyond the foresight of many, nearly four years ago.

The progression of concepts within the AI space  LLMs (Large Language Models), text generation, recognition and translation quickly transformed what can be achieved with chatbots.

In the last two years, generative AI advancements have shifted chatbots from using pre-defined templates to creating natural, engaging conversations.

Merging AI with chatbots, allowing them to evolve through data and feedback, has enhanced their performance progressively.

 

Transforming Chatbot Interactions

If you’ve experimented with PVA, whether casually or for enterprise purposes, you’re likely versed in the essentials of chatbot creation. These include:

• Trigger phrases to teach the bot different ways that someone might ask for a topic
• Topics to determine the general category of the conversation and narrow down the scope of conversation
• Actions to execute to the user’s request, such as providing information, completing a task or directing the user to further resources

Considering this, PVA’s move towards a context-aware, personalized chatbot was clearly foreseeable.

The Dawn of Copilot Studio

So what does all of this mean for the, now, Copilot Studio? Well, licensed users will be able to:

1. Quickly build, test, publish and analyse your own copilot experience by connecting to your websites, documents and business data within a single interface
2. Leverage plugins by importing or creating new ones from existing platform components. This can include data sources, connectors, flows, AI prompts and custom topics
3. Quickly generate topics (conversation paths) by using natural language prompts and receive a custom topic within a matter of seconds