Copilot Studio vs Power Virtual Agents - What Changed

April 10, 2026•8 min read•Michael Ridland

If you built chatbots on Microsoft Power Virtual Agents (PVA) over the past few years, you have probably noticed the platform no longer exists as a standalone product. Microsoft retired the PVA brand and rolled everything into Copilot Studio. The name changed, but so did quite a lot under the hood.

We have migrated several Australian businesses from PVA to Copilot Studio. Here is what actually changed, what got better, what got worse, and what you need to do about it.

The Short Version

Power Virtual Agents was Microsoft's low-code chatbot builder. It let non-developers create conversational bots using a visual authoring canvas with predefined topics, trigger phrases, and dialog flows.

Copilot Studio is the evolution of PVA. It keeps the visual authoring approach but adds generative AI capabilities, better integration with Microsoft 365 Copilot, and a more flexible architecture for building agents rather than simple chatbots.

The transition was not just a rebrand. It was a fundamental shift in how Microsoft thinks about conversational AI - from scripted bots to AI-powered agents.

What Changed Between PVA and Copilot Studio

Feature	Power Virtual Agents	Copilot Studio
Authoring approach	Topic-based with trigger phrases	Topics + generative AI answers
AI capabilities	Rule-based with some AI	GPT-powered generative responses
Knowledge sources	Manual topic creation	Auto-ingests from websites, SharePoint, files
Microsoft 365 integration	Basic	Deep Copilot ecosystem integration
Channel support	Web, Teams, Facebook, etc.	Same channels + Microsoft 365 Copilot
Pricing model	Per-session	Per-message with Copilot licensing
Extensibility	Power Automate flows	Power Automate + AI Builder + Azure AI
Agent orchestration	Not available	Multi-agent coordination support
Analytics	Basic conversation analytics	Enhanced analytics with AI insights
Name	Power Virtual Agents	Microsoft Copilot Studio

What Got Better

Generative AI Answers

This is the biggest improvement. In PVA, your bot could only answer questions you had explicitly configured as topics. If a user asked something you had not anticipated, the bot would fall back to a generic "I don't understand" response.

Copilot Studio agents can generate answers from your knowledge sources - SharePoint sites, uploaded documents, websites, Dataverse tables. You point the agent at your content and it can answer questions without you manually creating a topic for every possible query.

In our experience, this single change reduces the initial configuration effort by 60-70% compared to PVA. Instead of creating 200 topics, you point the agent at your knowledge base and then create specific topics only for workflows that need structured handling (like booking a meeting or submitting a form).

Knowledge Source Integration

PVA required you to manually create every piece of content the bot could reference. Copilot Studio can ingest content from multiple sources automatically.

Supported knowledge sources include:

SharePoint sites and document libraries
Public websites (crawled and indexed)
Uploaded files (PDF, Word, etc.)
Dataverse tables
Custom data through connectors

For Australian businesses with extensive SharePoint document libraries - which is most of our clients - this is a significant time saver.

Microsoft 365 Copilot Integration

Copilot Studio agents can be surfaced within the Microsoft 365 Copilot experience. This means employees can interact with your custom agents directly within their Copilot interface alongside Microsoft's built-in capabilities.

This was not possible with PVA. It operated as a standalone bot experience.

Agent Orchestration

Copilot Studio now supports scenarios where multiple agents work together. A front-door agent can route requests to specialised agents - one for IT support, another for HR queries, a third for customer service. PVA had no concept of multi-agent coordination.

Improved Analytics

The analytics in Copilot Studio give you better visibility into how agents are performing, which topics are triggering, where conversations are failing, and how generative answers are performing. PVA analytics were basic by comparison.

What Got Worse (or at Least More Complicated)

Pricing Complexity

PVA had a relatively simple per-session pricing model. You knew what a session cost and could estimate your monthly spend.

Copilot Studio's pricing is more complex. It ties into the broader Microsoft 365 Copilot licensing, uses a message-based consumption model, and the cost per interaction varies depending on whether you are using generative answers, premium connectors, or Azure AI services underneath.

We have seen Australian businesses struggle to accurately forecast their monthly Copilot Studio costs. If you want a detailed breakdown, we wrote about Copilot Studio pricing in Australia separately.

Migration Is Not Automatic

If you have existing PVA bots, they do not automatically become Copilot Studio agents. Microsoft provides migration tools, but in our experience, a straightforward migration works for simple bots. Complex bots with many topics, custom connectors, and Power Automate integrations often need significant rework.

For one client, we spent three weeks migrating a PVA bot with 150+ topics and 30 Power Automate flows to Copilot Studio. The migration tool handled about 60% of the topics correctly. The rest needed manual adjustment.

Learning Curve for Existing PVA Users

If your team was proficient with PVA, they will need time to learn the new Copilot Studio interface and concepts. The generative AI capabilities introduce new design patterns. Instead of thinking "what topics do I need to build," you now think "what knowledge sources do I point at, and which workflows need structured topics."

This is a better approach, but it requires a mindset shift.

Dependency on Microsoft 365 Ecosystem

PVA was relatively standalone. You could use it without deep Microsoft 365 integration. Copilot Studio is designed to work best within the Microsoft 365 ecosystem. If your organisation is not heavily invested in Microsoft, the platform's advantages diminish.

Migration Checklist - PVA to Copilot Studio

If you are planning to migrate, here is the process we follow with our clients.

Before Migration

Audit all existing PVA bots - document topics, flows, connectors, and channels
Identify which topics can be replaced by generative AI answers from knowledge sources
Review Power Automate flows for compatibility with Copilot Studio
Check premium connector licensing requirements
Estimate new message volumes under Copilot Studio pricing
Plan user communication about any changes in bot behaviour

During Migration

Use Microsoft's migration tool for initial conversion
Test every topic and flow manually - do not assume the migration tool handled everything
Configure knowledge sources for generative answers
Set up fallback behaviour for edge cases
Test across all deployment channels (Teams, web, etc.)
Validate analytics and reporting

After Migration

Monitor conversation quality for the first 30 days
Review generative answer accuracy and adjust knowledge sources
Optimise topics based on actual usage patterns
Train your team on the new authoring experience
Update internal documentation

Should You Migrate Now or Wait?

This depends on your situation.

Migrate now if:

Your PVA bots are due for a significant update anyway
You are already rolling out Microsoft 365 Copilot across your organisation
Your bots struggle with questions outside their configured topics
You want to reduce the maintenance burden of managing hundreds of topics

Wait if:

Your current PVA bots are working well and stable
You have no immediate plans to add Microsoft 365 Copilot
Your team does not have capacity for a migration project right now
Your bots are simple and the generative AI capabilities would not add much value

Microsoft has not announced a hard end-of-life date for PVA functionality, but the development focus is entirely on Copilot Studio. New features and improvements are only coming to Copilot Studio.

The Bigger Picture - Chatbots to Agents

The PVA-to-Copilot Studio transition reflects a broader industry shift. Simple Q&A chatbots are being replaced by AI agents that can reason, access multiple data sources, take actions, and coordinate with other agents.

If your current PVA bots are basically FAQ responders, the migration to Copilot Studio is an opportunity to rethink what your conversational AI can do. Instead of answering "what are your opening hours," your agents could be processing requests, looking up account information, scheduling appointments, and escalating complex issues - all within a single conversation.

We have seen Australian businesses use the migration as a catalyst to completely rethink their approach to conversational AI. One financial services client went from 5 PVA FAQ bots to 2 Copilot Studio agents that handle 3x the volume with better customer satisfaction scores.

Getting Help With Your Migration

Whether you are migrating from PVA or building new agents from scratch, Team 400 can help. As Copilot Studio consultants and Microsoft AI specialists, we have been through this transition with multiple Australian organisations.

We offer a migration assessment that gives you a clear picture of effort, cost, and timeline for your specific situation. Reach out to us to start the conversation, or learn more about our AI agent development services.