Innovate Your Business
with Our Insights!

Previously, archive jobs within the same scenario ran sequentially, meaning organizations with 50 to 200 or more legal entities had to wait for each entity's data to finish archiving before the next one could begin. For organizations holding years of General Ledger, Sales Order, or other high-volume transaction data, that process could take days or weeks to complete, creating pressure on SQL Server storage and delaying the movement of data into long-term retention in Dataverse. 

The new capability introduces a partition identifier called the Job Criteria Key, which organizations set when building an archive job contract. The scheduler uses this key to identify which jobs are targeting independent data sets and runs them simultaneously rather than queuing them. Microsoft has built in a zero-overlap guarantee, meaning jobs sharing the same scenario but carrying different criteria keys must process entirely distinct records, with no risk of conflict between concurrent jobs. 

For finance teams in large enterprises, the practical impact is a significant reduction in the time and resource cost of routine data housekeeping. Archiving that previously consumed a week of processing time could now complete in a fraction of that, depending on how many legal entities run in parallel. 

The self-healing capability is aimed directly at one of the most persistent pain points in robotic process automation, where desktop flows break when underlying applications change their interface or layout. Currently, broken flows require manual intervention to diagnose and repair. The new AI-powered approach would allow the system to detect and recover from those failures automatically, reducing the maintenance overhead that has made desktop automation difficult to scale in enterprise environments. 

The wave also introduces AI agent authoring for desktop flows, allowing makers to build automations using natural language instructions rather than recording or scripting each step manually. Microsoft is also adding Copilot Studio-powered actions inside cloud flows, deepening the connection between its automation and agent-building platforms. 

For organizations that have invested heavily in Power Automate as part of their process automation strategy, the self-healing capability could significantly change the economics of maintaining large desktop flow libraries. It addresses a problem that has limited enterprise confidence in unattended automation at scale. 

Process mining also reaches general availability in this wave, offering organizations a data-driven foundation for identifying where automation delivers the most value before building flows. The governance reporting consolidation announced alongside these updates suggests Microsoft is also responding to feedback from administrators managing large Power Platform environments. 

The 2026 release wave 1 plans reveal that Finance Agent will extend its capabilities inside Excel for data preparation and reconciliation tasks, and inside Outlook for customer communications that draw on financial data. The intent is to reduce the volume of time finance teams spend manually extracting, reformatting, and presenting ERP data to stakeholders. 

Microsoft is positioning Finance Agent as part of a broader strategy to make Dynamics 365 data accessible within Microsoft 365 rather than requiring users to operate inside the ERP directly. For finance teams that currently move between their ERP, spreadsheets, and email to complete routine tasks, the integration could reduce a significant amount of context switching. 

The announcement follows a pattern Microsoft has established across the Dynamics 365 suite, using Microsoft 365 Copilot as a unified front end for business application data. Finance Agent joins Sales Agent and the newly announced Service Agent in this architecture, each bringing domain-specific data and actions into the Copilot interface. 

Whether the approach delivers the productivity gains Microsoft is projecting will depend on how well the agent handles the complexity of real-world financial data, including entity structures, approval chains, and the highly customized configurations common in mid-to-large ERP deployments. The capabilities are expected between April and September 2026. 

The move positions Business Central as what Microsoft is calling an agentic ERP, a system where AI handles routine transactional work rather than simply assisting users who initiate it. The sales and purchase agents announced for the April to September 2026 release period will take on scenarios that currently require staff to enter, review, and process data themselves. 

Microsoft has also announced significant investment in developer tooling alongside these functional changes, including improvements to AL testing, debugging, Copilot extensibility, and agent design capabilities. The company framed this as a deliberate effort to support the partner and developer ecosystem as it builds on top of the new agentic infrastructure. 

The announcement comes as Microsoft faces growing pressure to demonstrate that its Copilot investments translate into measurable operational impact for mid-market businesses, which form the core of the Business Central customer base. Agentic automation, where AI takes action rather than offering suggestions, represents the clearest path to that outcome. 

The agent, now available in public preview, allows service professionals to manage cases, retrieve knowledge, update records, and trigger workflows entirely through natural language, without switching between their CRM, email, and internal tools. It draws on data from both Microsoft 365 applications including Outlook, Teams, and SharePoint, and Dynamics 365 Customer Service, giving agents a combined view of customer history and open cases from a single surface. 

Microsoft says the move is part of a broader shift toward making Copilot the primary system of engagement for business work, rather than a search or summarization tool sitting alongside existing applications. Service Agent can generate case summaries, flag priority items based on customer signals, retrieve relevant knowledge articles grounded in Dataverse and SharePoint, and add case notes or initiate child case creation, all without leaving the Copilot workspace. 

The product is designed to sit on top of existing Dynamics 365 Customer Service investments rather than replace them, with Microsoft describing it as an intelligent layer that brings information and actions to users across the application ecosystem. It carries enterprise-grade security and compliance alignment consistent with the broader Microsoft 365 Copilot framework. 

Microsoft has indicated that additional skills, workflows, and application support will be added as the public preview progresses. Organizations can enable the agent through the public preview documentation on Microsoft Learn.