Platform strategy starts with responsibility, not product logos.
- 01
Dynamics 365 and Power Apps run operational processes on Dataverse.
- 02
Fabric connects analytical workloads over OneLake, including engineering, warehouse, real-time and Power BI.
- 03
Foundry supports custom AI applications and agents with models, tools, evaluation and observability.
- 04
Semantics and ownership across the layers determine whether data improves work.
Dynamics 365 and Dataverse run the work process.
Customer, service and custom business records belong where employees perform the operational process. Dataverse provides the data model, roles, logic, APIs and ALM.
Not every piece of information should be copied into Dataverse. Systems of record and process handovers must be explicit.
- Dynamics 365 for standard processes
- Power Apps for custom processes
- Dataverse for operations and security
- Power Automate for workflow
Fabric organises data across workloads using OneLake.
Fabric combines ingestion, engineering, data science, real-time, warehouse, databases and Power BI in a SaaS platform.
The central question is which data products and metrics support which decisions and who owns quality, access and cost.
- Data Factory
- Lakehouse or warehouse
- Real-Time Intelligence
- Semantic models and Power BI
- Governance and catalog
Foundry turns models and data into controlled AI applications.
Foundry unifies models, agents and tools with identity, evaluation, tracing and monitoring.
It does not replace the operational process or data ownership. Agent tools need narrow permissions and observability.
- Model catalogue
- Prompt and hosted agents
- Tools and MCP
- Evaluation and tracing
- Entra identity and RBAC
Agent capabilities change roadmaps, not architecture discipline.
The 2026 release wave emphasises copilots and agents across Dynamics 365, including adaptive customer journeys.
Release plans still require evaluation by availability, region, licensing, privacy and operating readiness.
- Review release plans quarterly
- Separate preview from production
- Start with bounded tasks
- Define KPI, quality and fallback
Is your Microsoft platform clearly structured?
- 01
Which systems own which data?
- 02
Where do operations become analytics?
- 03
Are data products owned?
- 04
Are KPIs harmonised?
- 05
Which actions may agents perform?
- 06
How are ALM, security and cost governed?
- 07
How do releases enter the roadmap?
Get the briefing as a PDF
Receive the briefing directly and continue the discussion with NXTGN if useful. The direct download above remains available without a form.