The practical change is not that Copilot can answer questions about Power BI data for the first time. Microsoft has already been rolling out Power BI grounding in Microsoft 365 Copilot, allowing the assistant to answer from Power BI reports and semantic models. The Notebook integration is a different workflow: rather than asking Copilot to locate relevant analytics during a chat, a user will be able to deliberately attach a report to a persistent project workspace and use it with supporting documents.
That distinction gives teams a more controlled way to create recurring materials such as account reviews, operating updates, launch plans, and executive readouts. It also introduces several unanswered questions around data freshness, licensing, and report scope that Microsoft has not addressed in the roadmap entry.
A report becomes part of the notebook’s evidence set
Microsoft describes Copilot Notebooks as a workspace where Copilot is grounded in the references that a user has assembled for a particular project. Existing Microsoft support documentation says those references can include files, OneDrive and SharePoint content, folders, sites, and approved internal links. Copilot’s replies are then limited to material that has been added to the notebook and that the user is permitted to access.
The Power BI addition extends that model to a report. A finance lead could pair a quarterly-reporting Power BI dashboard with budget assumptions, meeting notes, and an existing PowerPoint deck; a sales manager could combine pipeline reporting with account plans and customer correspondence. In each case, the intended payoff is less context switching between a reporting portal and a drafting tool.
Microsoft’s roadmap says Copilot will use the added report data to produce more relevant summaries, presentations, and briefs. That wording matters. It promises grounding for generated work, but it does not promise that Copilot will create a fully auditable analyst-grade narrative, preserve every report filter, or identify every important exception in the underlying model.
Power BI reports are presentation layers over semantic models, often with filters, bookmarks, hidden pages, row-level security, and calculations that affect what a user sees. Microsoft’s current Power BI documentation already cautions that Copilot report summaries have practical limits: a report can contain hundreds of visuals, but Copilot summarizes only a subset of them, and very large table or matrix visuals may push the experience to use the semantic model rather than the report visual. Those limitations are documented for Copilot inside Power BI, not specifically for Notebooks, but they show why teams should test the new reference type against their most complicated reports before treating generated prose as a final management artifact.
This overlaps with, but does not replace, Power BI grounding in Copilot Chat
Microsoft’s Message Center announcement for Power BI integration says Microsoft 365 Copilot can answer questions using Power BI reports and semantic models, including cases where a user shares a report and cases where Copilot finds an appropriate report. That rollout began in Frontier environments in June 2026 and is scheduled to reach worldwide environments from late August through early September.
The Notebook roadmap entry lands during that broader rollout, but the two announcements describe different user experiences. The existing integration concerns discovery and answers in Microsoft 365 Copilot. Roadmap 569928 concerns explicit references inside Copilot Notebooks.
For IT teams, the difference is operational rather than cosmetic. A normal Copilot chat may be short-lived and oriented around a question; a notebook can become a reusable workspace that accumulates files, links, notes, and now reporting material. The durable notebook context is useful for long-running projects, but it means owners need to be more deliberate about who can access the notebook and its attached content.
Microsoft’s support documentation says recipients of a shared notebook are invited to linked files when the owner has permission to share them, while content that the recipient cannot access may still require an access request. The company also says Copilot uses only material added as a reference that the user is allowed to access. Those are important guardrails, but they do not eliminate the need to review sharing practices around commercially sensitive reports.
A report that is safe for a regional manager may not be appropriate for everyone participating in a cross-functional notebook. Existing Power BI permissions and row-level security remain central to that decision. Microsoft’s earlier Power BI grounding announcement says the service works within current security and permission models, but the new Notebook roadmap item does not explain how report-level access, saved filters, export permissions, or semantic-model permissions will appear in the notebook interface.
The roadmap leaves key behavior unspecified
The new item is unusually brief even by Microsoft 365 Roadmap standards. It does not say whether users will add a report by searching for it, pasting a link, selecting it from a Power BI picker, or attaching a static export. It does not say whether the notebook will reason over the report visuals, the associated semantic model, or both.
Most importantly, Microsoft does not say how data freshness will work for Power BI report references. Microsoft’s Notebook support documentation states generally that references stay current as data changes. But the roadmap entry does not specify whether a notebook attached to a Power BI report will query the current report on every prompt, cache a snapshot, respect refresh timing, or expose the report’s filter state to Copilot.
That is not a minor implementation detail. A notebook used for weekly operating reviews can become misleading if its supporting narrative reflects a prior refresh while users assume it reflects current numbers. Report owners should establish a simple validation step: compare key figures in a generated brief with the live report immediately before distribution.
Microsoft also has not disclosed whether all Power BI report types will qualify. The roadmap says “Power BI reports,” without mentioning paginated reports, Power BI apps, reports embedded in Teams, reports backed by specific Fabric capacities, or reports that rely on nonstandard visuals and custom calculations. It likewise says nothing about supported sovereign clouds; the stated scope is worldwide standard multi-tenant tenants only.
Licensing and capacity should be checked before rollout plans are made
Microsoft’s existing Notebook documentation draws a line between Microsoft 365 Copilot and Copilot Chat users. Microsoft 365 Copilot users can use premium notebook sources, including Teams, SharePoint, OneDrive content, and internal URLs, while Copilot Chat users have a more limited set of standard sources. The Power BI report reference is not yet listed in the public reference-type documentation, so Microsoft has not explicitly said which licensing tier will receive it.
Related Power BI grounding documentation provides a strong indication of the likely direction, but not a definitive answer for Notebooks. Microsoft’s Power BI integration announcement says a Microsoft 365 Copilot Premium license is required for that feature and that users still need the usual Power BI entitlement to view the Fabric or Power BI content. It also says users do not need Fabric Copilot access merely to use Power BI data in Microsoft 365 Copilot.
Organizations should therefore avoid assuming that a user who can open a Power BI report will automatically be able to add it to a Notebook. The September general-availability date is a target, not a published licensing matrix. Administrators should wait for the corresponding Microsoft 365 Message Center post or product documentation before changing license assignments or describing the feature as broadly available.
Notebook scale is another detail to watch. Microsoft currently says Microsoft 365 Copilot users can add more than 300 references, although only the first 300 are used for grounding; Copilot Chat users can add up to 50. Microsoft has not said whether a Power BI report will count as one reference, whether related semantic models count separately, or whether reports have different capacity limits.
A sensible pilot starts with narrow, well-understood reports
The best early candidates will be reports with clear ownership, predictable refreshes, and a small set of interpretable visuals. Teams should begin with a report whose numbers are already used in a recurring brief, then compare Copilot’s notebook output with the existing manually prepared version.
A practical pilot should include the following controls:
- Use a report with known row-level-security behavior and verify what separate notebook participants can retrieve through prompts.
- Record the report refresh time and validate every generated headline metric against the live Power BI report before sharing it.
- Test prompts against large tables, matrix visuals, hidden report pages, and filtered views to identify what Copilot overlooks or reframes.
- Keep the initial notebook audience small, because sharing a notebook can involve sharing access to linked references.
- Document the license combination used in the pilot, including Microsoft 365 Copilot and Power BI or Fabric entitlements.
Microsoft has put August 2026 Preview and September 2026 general availability targets on Roadmap 569928, but it has not yet published the details that determine whether this will be a straightforward reporting aid or another Copilot feature that requires careful tenant-by-tenant validation. For now, the consequential change is clear: Power BI reporting is moving from an adjacent Copilot data source into the notebook’s deliberate working context.
