Microsoft Opens Work IQ Preview for Dynamics 365 and Power Platform Data
Microsoft has opened a public preview of Work IQ for business applications, bringing a shared understanding of Dynamics 365 and Power Platform data to Microsoft Copilot and AI agents.
The technology is designed to help agents understand not only individual records but also how those records relate to one another within an organisation’s business processes.
The preview began on 30 September, with Microsoft’s rollout continuing through October.
For ERP teams, one of the most important aspects of Work IQ is where its understanding of the business comes from. Rather than creating an entirely separate knowledge model, the system builds its semantic understanding largely from configuration already present within Dynamics 365 and Power Apps.
That means the quality of the AI experience will depend, at least in part, on the quality of the underlying application configuration.
How Work IQ Understands Business Data
Work IQ creates a semantic model using information already held within business applications.
This can include:
- Views and forms
- Data relationships
- Business terminology
- Application configuration
- Defined business glossaries
The model provides a common interpretation of what records represent and how different pieces of information are connected.
That shared interpretation can then be used by multiple Copilot experiences and agents.
For ERP organisations, this creates an important dependency. If an application contains outdated views, poorly maintained relationships or inconsistent terminology, those problems can potentially affect how AI interprets the underlying business data.
Microsoft therefore gives administrators tools to refine the model. They can enable or disable specific signals, exclude outdated views or relationships and add business terminology that the system cannot determine automatically.
The objective is to create a cleaner semantic foundation for AI before agents begin using that information to support business processes.
Three Components Make Up Work IQ
Microsoft’s approach consists of three main elements: the semantic model, business skills and governed actions.
1. Semantic Model
The semantic model establishes the meaning and relationships behind business data.
It combines defined data scopes with semantic indexing and information drawn from existing applications and company terminology.
This gives agents a consistent understanding of the organisation’s data rather than requiring every individual agent to interpret the same information independently.
2. Business Skills
Business skills allow organisations to turn procedures into reusable instructions for AI agents.
A skill combines a business process with the context and resources an agent needs to carry it out.
Once defined, the skill can be reused across agents connected through the Work IQ Model Context Protocol (MCP) server, including agents created through Copilot Studio.
For ERP teams, this could provide a way to formalise processes such as renewals, procurement, customer service or financial workflows and make those procedures available to multiple AI experiences.
3. Governed Actions
The third component focuses on what agents are allowed to do.
Work IQ enables Copilot to recommend changes, route proposed actions through required approval processes and update source records within the permissions available to the user.
The approach is designed to keep AI actions connected to existing business controls rather than creating a completely separate permission structure.
Microsoft stores semantic models and business skills in Dataverse and has also introduced a new Dataverse Ask API to support interaction with the semantic layer.
A Cross-Application Example
Microsoft uses a sales renewal scenario to demonstrate how Work IQ can connect information across applications.
A sales opportunity may appear healthy in Dynamics 365 Sales. However, Work IQ can connect that opportunity to an unresolved customer service case in Dynamics 365 Customer Service.
It can also identify customer communications indicating that the service issue needs to be resolved before the renewal can proceed.
Copilot can then apply the organisation’s defined renewal skill and recommend the appropriate next steps.
The significance is that the AI is not simply answering a question about one record. It is connecting information across applications and interpreting those records within a broader business context.
This type of cross-application reasoning is likely to become increasingly important as organisations deploy AI agents across ERP, CRM and other line-of-business systems.
Who Controls Work IQ?
Microsoft divides administration across three different roles.
AI administrators manage AI access and policies through the Microsoft 365 admin centre.
Power Platform administrators determine which environments participate and control access within those environments.
Makers and developers determine how particular data is exposed within the boundaries established by administrators.
This division means that Work IQ governance is not controlled by a single team.
ERP, IT, security and application teams will therefore need to establish clear ownership of decisions around AI access, environments and data exposure.
Work IQ agents operate through Microsoft’s Agent 365 environment, which Microsoft says provides IT teams with visibility into how agents use business data.
Current Coverage and ERP Availability
Work IQ currently covers applications including Dynamics 365 Sales and Customer Service, Power Apps and selected third-party line-of-business applications.
Microsoft also plans to extend coverage to Dynamics 365 Finance and Operations, with the rollout scheduled for late October.
That timing is important for ERP teams.
Although Work IQ is already available in public preview, Finance and Operations support is still a planned expansion. Organisations looking specifically to apply the technology to financial, supply chain or other ERP processes will therefore need to account for the rollout schedule.
The service uses usage-based billing through Microsoft’s Copilot Credits model, meaning organisations will also need to consider consumption and cost management when evaluating potential deployments.
What This Means for ERP Teams
Clean up application configuration before enabling AI. Work IQ relies heavily on existing application structures, including views, forms and relationships. Removing obsolete configuration and maintaining business terminology can give AI agents a more reliable foundation.
Establish clear governance ownership. AI policies, environment access and data exposure are managed by different roles. Organisations should define who is responsible for each layer before expanding AI access across business applications.
Use cross-application scenarios to test the technology. Work IQ’s value lies partly in its ability to connect information across applications. Testing a scenario that spans CRM and customer service can help organisations assess data quality, governance and agent behaviour before extending the technology into more sensitive ERP processes.
Plan pilots around the rollout schedule. Finance and Operations support is expected later than the initial preview. ERP teams should therefore distinguish between capabilities that are available today and those that remain on the roadmap.
Include AI consumption in the business case. Usage-based billing means the cost of deploying AI agents will depend partly on how extensively they are used. Organisations should model potential consumption before moving from experimentation to production.
The Bigger Picture for Agentic ERP
Work IQ represents another step in Microsoft’s effort to create a common intelligence layer across its business applications.
The underlying idea is that AI agents become more useful when they share a consistent understanding of an organisation’s data, terminology and processes.
For ERP teams, that makes application configuration increasingly important.
Historically, maintaining views, relationships and business terminology was primarily an application-management task. With AI agents relying on those structures to interpret business information, configuration quality can now influence the quality of automated decision-making as well.
As Microsoft extends Work IQ into Finance and Operations, the technology could become an important component of its wider agentic ERP strategy, connecting business data, process knowledge and governed actions while keeping Dynamics 365 as the underlying system of record.