Microsoft announced the start of the Agent 365 Business Analytics rollout in Frontier on 17 November 2026. The tool brings together spending, usage and signals about the value of AI activity, covering Microsoft services and connected external services under the specified licensing conditions. It is relevant to businesses using several AI tools or agents that need to understand what consumes their budget, who uses the services and which business processes are associated with that activity.

Spending, usage and outcomes are different kinds of information

Microsoft describes Business Analytics as a tool built on the Agent 365 observability foundation. It brings spending, usage and value signals into one place. A service can generate spending without being used evenly across teams. Likewise, a large interaction count establishes activity, while interpreting that count depends on the work performed and on the definition of the outcome the business wants to examine.

The spending functions presented include identifying cost drivers and where expenditure is concentrated. Microsoft also mentions adoption barriers and employee surveys conducted from the application. Usage records and employee responses serve different purposes. The former describe recorded activity, while the latter can help explain how the tool is actually used within a team's workflow and where people encounter difficulty or need a different process.

To assess value, Microsoft lists signals such as assisted hours, assisted value, success rate and satisfaction. An indicator's label matters to its interpretation. Estimated assisted hours are not the same record as billed hours or savings confirmed in financial accounts. For a business analysis, these values need to be read against their definition, the period observed and the operations included in that particular calculation.

Levels of analysis for an AI investment

  1. Consumption

    Users, services, agents and models

  2. Cost

    Recorded spending and budget limits

  3. Activity

    Use cases and tasks performed

  4. Assessment

    Defined indicators, forecasts and decisions

Overhead view of a laptop and printed charts on a desk
Nataliya Vaitkevich

Forecasts and scenarios for AI investment

Business Analytics includes spending forecasts and scenario planning. These connect consumption information with investment decisions: which service uses the budget, how usage changes and which activities are associated with that spending. A forecast is a calculation for a future period based on its data and assumptions. It remains separate from costs already recorded and from the final outcome of the company's investment decision.

Microsoft mentions exploring insights in natural language and sharing views through Outlook or Teams. This connects analysis with discussions between people responsible for technology and people responsible for business processes. A report about a support agent can be interpreted differently from one about preparing quotations, because the work, success criteria and review of the output differ between those processes and the teams operating them.

One possible analysis structure, without assumed financial values, links the service, the team using it, the task performed and the cost for the period. Consistent fields also help compare months and identify volume changes. Such comparisons become meaningful when definitions and periods remain consistent. Adding percentages or estimates from unrelated systems does not, by itself, establish a common measure of business performance.

Cost Management retains a separate operational role

The announcement separately presents Agent 365 Cost Management and its consumption controls. Spending can be tracked across users, services, agents and models. Microsoft describes limits and alerts at user, group, organisation and resource level. These controls address budget administration and the allocation of responsibility, whereas value analysis examines the relationship between expenditure, activity and the outcome of the underlying business process.

Microsoft also includes controls over the models teams can use, allowing the level of model to be matched with the work. This is an administration function, rather than a universal ranking of models. A company's task classification might distinguish internal summaries, field extraction and more complex operations, each with different requirements for data, review and access to tools and business systems.

In its Dynamics 365 material from the same event, Microsoft places some additions in December 2026, including value insights across Microsoft and external services and delegation to business leaders. That timetable must remain separate from available functions and the Frontier rollout beginning in November. A shared list of announcements does not mean that every capability has the same availability stage in every organisation or tenant.

In this context, observability means information about an agent's identity, activity and tools, where the connected platform supplies it. A cost associated with an identified agent and a period can be traced to the relevant workflow. This connects technical operations with business responsibility while keeping consumption measurement separate from interpretation of the benefits.

What each indicator shows and how to read it

InformationWhat it describesUseful distinction
SpendingCost for a periodSeparate from a future forecast
UsageRecorded activitySeparate from business outcomes
Assisted hoursAn indicator of AI assistanceSeparate from billed hours or accounting savings
Success rateThe defined outcome of a taskRetain its definition and observed population

Licensing and connections determine analytical coverage

According to the 17 November announcement, Business Analytics begins rolling out in Frontier. Analytics for Microsoft services require at least one Microsoft 365 Copilot licence in the tenant. Analytics for connected non-Microsoft AI services require Agent 365 and/or ME7 licensing. These are the access conditions stated in the announcement. Having an ordinary Microsoft subscription does not automatically establish coverage of every external data source or AI service.

Microsoft describes connecting AI spending to existing reporting through Microsoft Graph APIs, data exports and Fabric connectors. These mechanisms can bring AI information into budgeting and chargeback workflows. For business dashboards and automated reports, available fields, update frequency and access rights need to match the company's report structure. An export supplies data; interpreting it still depends on the reporting model and the definitions attached to its fields.

For entrepreneurs, the most direct application is in businesses already using Microsoft infrastructure or several connected AI services. When designing AI business automation, consumption records, ownership and task outcomes can be defined from the outset. Agent 365 adds concrete functions to that administration. Its value for a particular company is assessed from that company's activity and costs, retaining the indicator definitions in the report rather than replacing them with general assumptions.