Understand AI assisted hours and AI value (Frontier)

Important

This feature is part of the Frontier preview program. Frontier connects you directly with Microsoft's latest AI innovations. Frontier previews are subject to the existing preview terms of your customer agreements. As these features are still in development, their availability and capabilities might change over time.

Applies to: Agent 365 Business Analytics, Microsoft 365 Admin Center

AI assisted hours and AI value help organizations understand the economic value of the work that AI completes on behalf of users. These metrics translate AI-assisted work into equivalent human work time and its estimated monetary value.

By measuring the work AI contributes, organizations can better understand how AI supports productivity, identify valuable use cases, and evaluate the potential economic benefits of their AI investments.

Important

This entire article applies to Cowork assisted hours and value, including customization. It does not apply to Copilot assisted hours and Copilot assisted value, including customization.

What are AI assisted hours?

AI assisted hours estimate the amount of time a person would have needed to complete the work performed by AI.

The metric considers how frequently AI assists with a particular use case, whether the requested work was completed, how much of the work AI can perform, and how long a person would typically need to complete that work without AI.

For example, AI might help employees prepare business plans, analyze information, or summarize documents. Assisted hours estimates the equivalent human effort associated with these contributions.

Unlike traditional usage metrics, which measure how frequently AI is used, assisted hours estimate the amount of work AI contributes.

Important

AI assisted hours represent estimated equivalent human work, not actual employee time saved. They don't necessarily indicate that employees spent fewer hours working.

What is AI value?

AI value translates AI assisted hours into an estimated monetary value using an hourly labor rate.

The metric provides a standardized way to understand the potential economic contribution of AI-assisted work.

For example, if AI contributes work equivalent to 10 human hours and the estimated hourly rate is $50, the corresponding AI value is $500.

Important

AI value is an estimate of the labor value associated with AI's contribution. It doesn't represent realized cost savings, increased revenue, or a guaranteed return on investment.

Use this resource to learn more about the methodology behind the calculations described below.

How AI assisted hours are calculated

AI assisted hours are calculated using four components:

AI assisted hours = Use case frequency × Completion rate × AI contribution × Time per use case

Each component represents a different aspect of AI-assisted work.

Component Description
Use case frequency Number of times the use case was identified
Completion rate Percentage of identified use cases where AI completed the requested task
AI contribution Maximum percentage of the use case that AI can complete
Time per use case Estimated hours a person would need to complete the use case without AI, based on Microsoft Research estimates. This value can be customized by an administrator.

Example calculation

Consider an AI use case with the following characteristics:

Component Example
Use case frequency 1,000
Completion rate 80 percent
AI contribution 75 percent
Time per use case 0.5 hours

1,000 × 80% × 75% × 0.5 = 300 hours

In this example, AI contributed work equivalent to an estimated 300 hours of human effort. The calculation accounts for both the number of times the use case occurred and the proportion of work AI successfully contributed.

How AI value is calculated

AI value converts assisted hours into an estimated monetary value.

AI value = AI assisted hours × Hourly rate

Component Description
AI assisted hours Estimated human work time equivalent to the work contributed by AI.
Hourly rate Estimated hourly labor cost used to assign a monetary value to AI-assisted work. This value can be customized by an administrator.

Example calculation

Continuing the previous example:

Component Example
AI assisted hours 300
Hourly rate $60
AI value $18,000

300 hours × $60 = $18,000

This represents an estimated $18,000 in equivalent human labor value associated with AI-assisted work. It doesn't mean the organization reduced its operating expenses by $18,000.

How Microsoft estimates assisted hours and value

Microsoft uses a research-based methodology to estimate the work AI contributes. The methodology combines task classification, estimated human effort, AI capabilities, and task completion.

Identifying use cases

Microsoft uses a standardized taxonomy containing more than 6,300 business-process tasks relevant to information work. This taxonomy helps identify the types of work performed with AI and provides a consistent framework for measuring productivity contributions across different use cases.

Estimating human effort

For each use case, Microsoft estimates how long a skilled person would typically need to complete the work without AI assistance. These estimates are developed using AI models that evaluate representative task scenarios and validated using human evaluators.

Multiple scenarios are considered to account for differences in task complexity and execution requirements. Median estimates help reduce the influence of unusually simple or complex scenarios. The resulting estimates provide the baseline for the Time per successful use case component of the calculation.

Estimating AI contribution

Not every part of a use case can necessarily be completed using AI. For example, AI might assist with preparing a business recommendation, but human judgment might still be required to validate findings, make decisions, or approve the final recommendation.

Microsoft estimates the maximum proportion of each use case that AI can perform. This percentage establishes an upper limit on the work attributed to AI.

Accounting for task completion

The calculation also accounts for whether AI completed the requested task. The completion rate represents the percentage of identified use cases where AI successfully completed the requested work.

Microsoft estimates the completion rate by evaluating the last turn of each conversation. If that turn is successful, the conversation is marked as completed. The completion rate for a use case is the average across all conversations where that use case was identified.

Including completion rate helps ensure that estimated assisted hours reflect successful AI-assisted work rather than simply counting every identified use case.

Estimating monetary value

Microsoft associates tasks with occupational work activities from the O*NET framework. The methodology uses average compensation for occupations associated with those activities to estimate an hourly labor cost.

Annual compensation is converted to an hourly rate using a standard 2,080-hour work year. This provides a consistent basis for translating AI-assisted work into estimated monetary value.

Explore the hours and value calculator

The hours and value calculator provides transparency into how AI assisted hours and AI value are estimated. You can explore the calculations for individual use cases and understand the factors contributing to the reported metrics.

The calculator provides two views.

Hours calculator

The Hours calculator displays the components used to estimate assisted hours for each use case:

  • Use case frequency

  • Completion rate

  • AI contribution

  • Time per use case

  • Calculated AI assisted hours

Use this view to understand how frequently a use case occurs and how its estimated contribution translates into equivalent human work time.

Value calculator

The Value calculator displays the components used to estimate the monetary value of AI-assisted work:

  • AI assisted hours

  • Hourly rate

  • Calculated AI value

Use this view to understand how assisted hours translate into estimated labor value.

Find specific use cases

The calculator organizes use cases into categories and provides a search field to help you locate specific categories or use cases. Each row displays the calculation inputs and the resulting estimate for that use case. This breakdown helps explain why different use cases contribute different amounts of assisted hours or value.

Customize calculation assumptions

The calculator uses Microsoft Research estimates to provide a starting point for measuring AI-assisted work. Administrators can customize certain assumptions to better reflect their organization's circumstances.

The following values are customizable:

Setting Description
Time per use case Estimated time a person would need to complete the use case without AI. Administrators can adjust this estimate to reflect their organization's workflows and expected effort.
Hourly rate Estimated hourly labor cost used to calculate AI value. Administrators can adjust this value to reflect their organization's labor-cost assumptions.

Customizing these assumptions allows organizations to tailor the estimates to their business context.

For example, an organization might determine that a particular use case typically requires more human effort than the default estimate. Adjusting the time estimate changes the assisted hours attributed to that use case.

Similarly, adjusting the hourly rate changes the estimated monetary value without changing the underlying assisted hours.

Note

AI assisted hours and AI value are estimates based on the calculation inputs. Changes to administrator-configurable assumptions can affect reported values.

How to interpret assisted hours and value

AI assisted hours and AI value provide complementary perspectives on AI’s productivity contribution.

Metric What it measures What it doesn't measure
AI assisted hours Estimated equivalent human work contributed by AI Actual employee hours saved
AI value Estimated monetary value of AI-assisted work Realized financial savings
AI usage How frequently AI is used The amount or value of work completed
  • Understand AI's productivity contribution. Estimate the volume of work supported by AI across the organization.

  • Identify valuable use cases. Understand which types of AI-assisted work contribute the most estimated value.

  • Track trends over time. Monitor how AI's estimated contribution changes as adoption and usage evolve.

  • Inform investment decisions. Consider estimated AI value alongside usage, costs, and other business outcomes.

  • Understand calculation assumptions. Use the calculator to examine how use case frequency, completion, AI contribution, and human effort estimates influence results.

Important considerations

These metrics are estimates. Calculations rely on modeled human task duration, AI capability, and task completion rather than directly observing how long an employee would have taken to perform the work.

Assisted hours aren't equivalent to time saved. AI might allow employees to complete additional work, improve existing work, or perform tasks they otherwise wouldn't have undertaken. These benefits don't necessarily translate into a corresponding reduction in working hours.

AI value isn't realized financial savings. The metric estimates the labor value associated with AI's contribution. It doesn't establish that an organization reduced labor expenses by the same amount.

Different use cases contribute different amounts of value. Use cases vary in frequency, completion rate, AI contribution, and estimated human effort. As a result, two use cases with similar usage volumes can produce different estimates.

Customized assumptions affect results. Administrator-configured time estimates and hourly rates can influence reported assisted hours and value.

Estimates might evolve. Improvements in task classification, AI capabilities, and measurement methodologies can affect calculated values over time.

Frequently asked questions

Does one assisted hour mean an employee saved one hour?

No. One assisted hour represents an estimate of work equivalent to one hour of human effort. It doesn't necessarily mean an employee spent one fewer hour working.

Why can two use cases with the same frequency generate different assisted hours?

The use cases might have different completion rates, AI contribution percentages, or estimated human task durations. Assisted hours account for these differences rather than assigning the same value to every use case.

Why is AI contribution less than 100 percent?

Some use cases require human participation, judgment, or actions that AI can't perform. AI contribution accounts for these limitations by estimating the maximum percentage of the use case that AI can complete.

Can I change the assumptions used to calculate assisted hours and value?

Administrators can customize the estimated time per successful use case and hourly rate. These settings allow organizations to adjust calculations to reflect their own workflows and labor-cost assumptions. Learn more.

Does AI value represent actual cost savings?

No. AI value represents an estimate of the monetary value of AI-assisted work. Realized savings depend on organizational processes, labor costs, and how AI-assisted work affects business outcomes.

Can assisted hours and AI value be used to calculate ROI?

These metrics can inform an ROI analysis, but they don't independently establish ROI. A complete ROI assessment should consider AI-related costs, realized business outcomes, and other relevant benefits or expenses.