FAQ for reconciliation with Finance agents (preview)

[This article is prerelease documentation and is subject to change.]

These frequently asked questions (FAQ) describe the AI impact of Finance agents features in Excel.

What are Finance agents in Excel?

Finance agents provide a sidecar experience for accountants and other finance professionals in Excel. This experience transforms the way that these users perform their core tasks. They can complete business processes without having to sign in to the financial system. Instead, they can stay in the context of Microsoft 365 productivity tools such as Excel. Thanks to recommended actions and summarization, users can complete tasks more efficiently and accurately. Therefore, they can respond more quickly to changing conditions.

What can Finance agents in Excel do?

The list of capabilities that are offered is growing. Currently, Finance agents provide an AI-powered reconciliation experience. It can optimize the financial reconciliation process by reducing the manual effort that's required to run the process in Excel and eliminating the need to switch context for the specific job that must be done. For finance professionals who are engaged in reconciliation, Finance agents provide tools that speed up the analysis of multiple Excel-based data sources to identify matched, potentially matched, and unmatched transactions.

What is the system's intended use?

As an AI assistant to finance professionals, the system provides a set of intended uses for the reconciliation experience in Excel:

• Determination of reconciliation process parameters – With the help of AI, Finance agents use the selected reconciliation data sets to analyze column names and types of data. Based on this analysis, it determines and suggests the best reconciliation vectors and also columns that contain values that must be reconciled.
• Classification of financial data as unmatched, potentially matched, and perfectly matched – Finance agents run the reconciliation process according to selected parameters to determine any type of data discrepancy.
• Authoring reconciliation report with report summary – Finance agents create a report that lists three groups of classified transactions (unmatched, potentially matched, and matched) together with a generative AI–provided summary of the report.

How were Finance agents in Excel evaluated? What metrics are used to measure performance?

We evaluate features both qualitatively and quantitatively. To assess the quality of the feature, we are conducting user studies with finance professionals to gather their feedback on their experiences, thoughts about the feature's quality, and suggestions for improvement. Additionally, we are monitoring telemetry data to track the number of finance professionals who tried the feature, the success rate of the feature, and the ratio of positive to negative feedback. Before we released Finance agents in preview, we conducted extensive testing to ensure its functionality. If you encounter any issues with the content that's generated, or with any of the listed use cases, provide feedback. Your feedback is used to enhance Microsoft's products and services. Your organization's IT admins might have access to your feedback data for management purposes. For more information, read the Privacy Statement.

What are the limitations of Finance agents in Excel? How can users minimize the impact of Finance agents in Excel's limitations when they use the system?

To use this capability, your environment must be in the US region. Preview features aren't meant for production use and might have restricted functionality. These features are available before an official release, so that customers can get early access and provide feedback. This capability might be subject to usage limits or capacity throttling. In terms of functionality, data for reconciliation must be brought to Excel as two structured tables to start the reconciliation process. Reconciliation parameter suggestion doesn't use advanced data analytics techniques to propose more adequate reconciliation vectors. The ability to generate detailed instructions for the best way to address inconsistencies is limited by a lack of business context for grounding or more context for transactions that are reconciled as part of the inputs for the language model prompts. In Excel, trailing digits are changed to zeros when long numbers are used in cells. This behavior might lead to false positives when transactions are matched.

What operational factors and settings allow for effective and responsible use of Finance agents in Excel?

Follow these steps to make the most of the feature.

As an admin, you can enable and have full control over add-in deployment.

As an app user, you receive suggestions for reconciliation vectors. You can choose to review, accept, or override these suggestions. As a result of the reconciliation process, you get a report that includes classified transactions. Generative AI provides a report summary that can be regenerated before it's included in the report. The more data that's available for analysis, the more accurate the suggestions that are provided in the summary. Financial data for analysis should always be provided in two structured data sets in Excel that are formatted as tables. Each table must have at least two columns, one of which will be used as a mapping key (reconcile by) and the other of which is numeric (typically, monetary), so that values can be compared. The provided data and the selection of reconciliation vectors determine the classification of transactions in the report and therefore affect the report summary. Acceptance of the reconciliation criteria without adequate scrutiny might lead to incorrectly classified transactions and therefore an inaccurate summary. Use of the reconciliation tool outside the financial ledger-to-subledger reconciliation, accounts payable (AP) reconciliation, or accounts receivable (AR) reconciliation is considered misuse during preview.