Enhancements to the Outlook Add-in
One of the key areas of improvement in Copilot for Finance is the investment in the Outlook add-in, designed to enhance the collections experience for finance professionals and those handling accounts receivable communications.
The Copilot for Finance Outlook add-in leverages generative AI to help users summarize and compose emails, providing instant context by integrating with an organization’s ERP system. Key features include:
- Summarizing customer interactions and outstanding balances.
- Attaching important documents such as invoices and account statements.
- Logging correspondence and notes for improved transparency.
- Directly updating the ERP system with the latest invoice dispositions and customer details.
Expanded Backend Integration Support
To enhance usability, Copilot for Finance has introduced significant backend improvements:
- Expanded ERP Support – Now compatible with Business Central, SAP, and any ERP system that supports REST APIs and OAuth authentication via a custom connector.
- Demo Mode – Users can now evaluate the add-in without ERP integration, allowing for easier product testing and exploration.
- Improved Email Summarization – Email summaries now include ERP-referenced account balances, invoice due dates, and amounts whenever invoice numbers appear in email threads.
Advancements in Data Preparation & Reconciliation
A new feature set in Copilot for Finance focuses on data preparation and reconciliation within Excel. Initially, Copilot addressed three of the six core reconciliation steps: matching, classifying, and reporting. The goal is to support the entire reconciliation process, moving from an assistive solution to a more automated, agent-driven approach.
With the addition of data preparation capabilities, the tool now moves beyond simple reconciliation to support the validation and preparation steps as well.
Upcoming Reconciliation Improvements
To further streamline reconciliation, upcoming improvements include:
- Retrieving source data from multiple files instead of requiring all data within a single Excel workbook.
- Defining multiple matching rules to categorize records as matched or potentially matched, providing greater flexibility in custom reconciliation logic.
- Enhanced reporting with additional insights, including:
- Reconciled and unreconciled item attribution for greater transparency.
- Drill-back functionality to trace reconciled items to their original source data.
- Appending source data details to provide full audit trails in reconciliation reports.
- Customizable report groupings and configurations, offering users more control over reconciliation summaries and contextual data references.
Expansion of Reconciliation Capabilities
Further planned enhancements include:
- Single-source reconciliation for scenarios where only one data source is required.
- Support for three or more data sources, allowing for complex multi-source reconciliation processes.
- Automation and AI-driven reconciliation, leveraging agent capabilities to streamline workflows and minimize manual intervention.
Variance Analysis
Variance analysis is a crucial process in financial planning and control, helping organizations assess performance across various dimensions. Finance professionals often spend significant time working in Excel, using traditional manual or semi-automated methods to perform variance analysis. However, these conventional approaches present multiple challenges.
Challenges in Variance Analysis
- Time-Consuming Processes – Identifying and calculating variances across multiple lines, periods, and dimensions is labor-intensive. Manual operations are also prone to errors, increasing the risk of inaccurate data interpretation.
- Limited Analytical Depth – Standard variance analysis typically identifies variances but does not offer deeper insights into root causes or patterns without significant additional effort.
- Scalability Issues – As businesses grow, managing and analyzing large datasets becomes cumbersome. Excel’s native capabilities may not be sufficient to handle the increasing complexity, limiting the ability to perform timely variance analysis.
To address these challenges, new variance analysis capabilities are being developed to work with external data sources, allowing analysis of data at rest without the need for importing it into Excel.
Objective of Enhanced Variance Analysis
The primary goal of enhanced variance analysis is to reduce the time required for analysis and improve accuracy in summarizing findings. This minimizes the risk of omissions and inaccuracies, allowing finance teams to focus on understanding the ‘why’ behind the data rather than spending time on manual tasks.
Data Handling in Variance Analysis
Diagnostic Data Logging
For diagnostic purposes, metadata about usage will be logged, including execution times and scenario success rates. However, user-entered input, prompts, ERP system data, or Excel sheet contents will not be logged. Standard Microsoft data handling policies apply to this telemetry, ensuring data security and compliance.
Data Sent to Large Language Models (LLMs)
- For email summarization, only email content and user prompts are sent.
- For Excel-based scenarios (e.g., data reconciliation, variance analysis), user prompts, column headings, and a subset of Excel data are included.
- None of this data is stored after being sent to the LLM. It is not used for model training and is securely processed within Azure OpenAI.
Data Persistence and Processing
- Persistence – Currently, Copilot for Finance does not persist data unless users intentionally save it.
- Collections Use Case – Users can save notes and emails to a Dataverse environment.
- Future Enhancements – Additional persistence for Excel scenarios may be introduced, particularly for automation, but always with user-driven actions ensuring transparency.
- Processing Location – All data processing occurs within the user’s tenant’s geographic boundary on Microsoft’s segregated servers, ensuring no risk of data leakage between customers.
Future Developments in Variance Analysis
- Integration with Multiple Data Sources – Expanding beyond single Excel workbooks to support data retrieval from multiple files.
- Flexible Matching Rules – Users will be able to define multiple matching rules for categorization.
- Enhanced Reporting – Improved reconciliation reports with deeper attribution and drill-back capabilities.
- Automation and AI Agents – AI-driven automation for variance analysis will further streamline financial workflows.