Dataverse data is coming to Microsoft 365 Copilot’s Work IQ – is your business ready?

What makes Dataverse intelligence tick, and why should it make organisations think differently about their business data?

Headshot Morgan A

Morgan A

|

Senior Solution Developer

Posted on: May 29, 2026

|

Last updated: June 9, 2026

|

13 min read

Headshot Morgan A

Morgan A

|

Senior Solution Developer

Posted on: May 29, 2026

|

Last updated: June 9, 2026

Read time: 13 minutes

This June, you’ll be able to talk to Microsoft 365 Copilot about the business data sitting behind your Dynamics 365 and model-driven Power Apps. Not by building an advanced custom agent, or by having to open yet another tab or app – just by asking Copilot in whichever Microsoft app you’re already working in.

The feature behind this is called Dataverse intelligence, and it’s about to move into public preview. It extends Work IQ – the context layer behind Microsoft 365 Copilot’s ‘Work‘ (now ‘Work IQ’) mode that grounds it in your organisation’s data – with critical new capabilities that move Dataverse beyond its familiar role as the database for business applications, and position it firmly as Microsoft’s agent data platform.

In this blog, we’ll explore what makes Dataverse intelligence tick, and why it should make organisations think differently about their business data – and why your security model deserves a fresh look before it goes live.

What is Dataverse intelligence?

Currently, Copilot is fantastic with organisational context: it knows what’s been said, shared, and scheduled across your Microsoft 365 tenant. But that context is still largely made up of the outputs of work – not the business records, rules, and workflows that explain why the work is happening in the first place.

Dataverse intelligence is what closes that gap, through two core capabilities.

The first is making Dataverse data available to Microsoft 365 Copilot – this is the headline item, and the engine behind it is a fundamentally reworked Dataverse Search. Microsoft has rebuilt it as something closer to a conversational reasoning layer over live business data than a traditional search tool. Instead of expecting users to know which table to query or how records relate to one another, the new agentic architecture figures that out itself – planning how to answer the question, navigating relationships across tables, and self-correcting when the results don’t look right. It supports multi-turn conversations so follow-up questions keep their context, glossaries for business-specific language, and learning from usage patterns over time – so it gets sharper the more your team uses it.

The second is business skills – reusable, natural-language instructions that capture and standardise your team’s processes, policies, and ways of working, giving agents a consistent basis for how work should be handled across the organisation. Skills are defined once, stored centrally in Dataverse, and managed through a dedicated interface in the Power Apps maker portal – so when a process changes, you update it in one place, and every agent in your business picks up the change automatically.

Business skills dataverse inforgraphic

That’s not limited to Microsoft 365 Copilot, either – the Dataverse MCP server is also getting a significant upgrade as part of this wave. If you’re not familiar with MCP – Model Context Protocol – think of it like USB-C for AI agents: a single, standard connection that lets any compatible agent plug into a data source and actually interact with its data, rather than just talk about it. With this release, agents connecting through MCP can automatically discover and use those same business skills – so whether they’re built in Copilot Studio, Microsoft Foundry, GitHub Copilot, or any other MCP-compatible platform, they work from the same operational knowledge.

Model Context Protocol

The security question you need to be asking

At the heart of this release lies a security dimension that organisations can’t afford to ignore.

Some businesses, often without meaning to, have ended up with something closer to ‘security by obscurity’ than real access control – data is technically accessible, but the UI makes it hard to find. Permissions may be broader than they should be, but because users don’t know exactly where the data lives – or aren’t technically savvy enough to dig it up for themselves – the risk can stay hidden for years.

In an AI-enabled workplace, that falls apart. If a user has access to the data – even access nobody realised they had – all they need to do is ask the right question.

To be clear, Copilot fully respects your existing security model. It will only ever return data a user already has permission to see. But that’s exactly the problem – and we’ve already seen this play out with SharePoint.

When SharePoint oversharing first hit the headlines, many organisations didn’t blame their permissions – they blamed Copilot. But the real issue was that nobody had fully understood just how much access users actually had until Copilot made it trivially easy to use.

Dataverse brings that same reckoning into a more operational space – because if Copilot and agents can work with your business records, then every gap in your security role design, every overly broad permission, and every sensitive field that was only ‘safe’ because the app didn’t surface it is now just one plain-English question away from being exposed.

What you should do next

The direction from Microsoft is clear: Dataverse is no longer just a database for the Power Platform. It’s becoming the bridge between your business systems and Microsoft’s AI stack – and the organisations best-placed to benefit will be the ones whose key business data is structured, governed, and living in Dataverse.

That doesn’t mean you need to migrate everything overnight. But if you’ve been weighing up whether to move a business process into a model-driven app, or evaluating if Dataverse is worth the investment for a new project, there’s never been a better time. If the data matters enough to run your business, it matters enough for Copilot and agents to understand – and you should be asking if you can afford for them not to.

And with public preview planned from June 2026 across Microsoft 365 Copilot in Teams, Outlook, Word, Excel, PowerPoint, and Copilot Chat, this is the time to think seriously about how much you trust your current Dataverse estate. Is your data model clear enough for Copilot and agents to reason over, or does it depend on knowledge that only your makers and admins really understand? If every licensed user in your organisation could query your business data in plain English tomorrow, are you confident in what they’d find?

Preview does not mean every scenario will be production-ready on day one. But ready or not, business data is moving into the everyday Copilot experience, and organisations need to be prepared for what that exposes – both the opportunities and the risks.

We’ll certainly be digging into what Dataverse intelligence can do as it rolls into preview, and we’d encourage any organisation running Dynamics 365 or Power Platform to do the same. If you want to understand what this could mean for your organisation – from exploring how Copilot could work with your Dataverse data, to tightening up your data model and security – get in touch. We’d be happy to talk it through.

Last updated 9 Jun 2026

About the Author: Morgan A

Morgan A
Morgan is a Senior Solution Developer who works with Intelogy's Business Applications & Insights team. She enjoys solving complex problems and helping clients turn business challenges into practical, user-centric solutions that improve the way their teams work day to day.

Table of contents

Go to Top