As Microsoft continues to advance Dynamics 365 Business Central as a leading cloud ERP platform for small and mid-sized organisations, one of its most significant recent innovations is the introduction of Agentic AI. These autonomous capabilities enable the system to interpret information, execute multi-step workflows, make context-aware decisions, and interact with users much like a trained professional.
This article provides a focused, business-oriented review of Business Central’s two key agentic AI capabilities – the Sales Order Agent and the Payables Agent – highlighting the operational challenges they address and how they enhance accuracy, speed, and efficiency across sales and finance functions.
Understanding Agentic AI in Business Central
Agentic AI represents a shift from traditional AI models that simply provide suggestions or predictions. Instead, agentic AI is designed to act. These agents are capable of navigating Business Central’s user interface, reading and interpreting information, executing transactions, validating data, and asking for human input when necessary.
Several characteristics differentiate agentic AI from earlier automation tools:
Autonomous Execution
Agents can independently perform tasks such as creating documents, selecting data, applying business rules, and drafting responses. They simulate the behaviour of a trained human user, using natural-language instructions to guide actions.
Contextual Intelligence
Agents operate with an understanding of business context. For example, when creating a sales order, the agent considers item availability, customer history, previous orders, committed quantities, and delivery timelines.
Auditability and Governance
Every step executed by the agent is logged. Users can review what the agent did, why it did it, and make corrections where needed. This ensures transparency, traceability, and compliance with internal control frameworks.
Human Oversight by Design
Agents proactively involve users when confidence in a decision is low – for instance, when a vendor cannot be identified from an invoice, or when requested items appear ambiguous.
Agentic AI, therefore, enables automation that is intelligent, safe, and aligned with organisational controls.
Sales Order Agent in Business Central
The Sales Order Agent is the first end-to-end autonomous agent introduced in Business Central. It automates the entire “email-to-order” process, commonly found in customer service, inside sales, and order fulfilment functions.
What the Sales Order Agent Does
Email Intake and Interpretation
The agent monitors designated mailboxes for customer requests. It analyses the content of emails – both text and attachments such as PDFs or images – to identify order intent, item references, quantities, delivery dates, and other relevant information.
Customer and Document Identification
The agent matches email senders to existing customers and aligns incoming messages to the correct quote or order, even when the customer does not explicitly reference a document number.
Intelligent Product Identification
Many customer emails reference items informally (e.g., “the blue model we ordered last time”). The agent searches across item descriptions, attributes, translations, and historical orders to determine the correct item.
Availability and Delivery Calculation
If stock is insufficient, the agent uses Business Central’s capable-to-promise (CTP) logic to calculate realistic delivery timelines.
Quote Drafting and Updates
The agent drafts sales quotes with all necessary details – pricing, quantities, delivery dates, external document references – and revises them as customers send follow-up emails with changes.
Conversion to Sales Orders and Email Drafting
Once the customer approves, the agent converts the quote to a sales order and prepares email responses with attachments for human review before sending.
The outcome is a streamlined, consistent, and significantly faster order intake process, especially beneficial for organisations with high email-driven sales volumes.
Payables Agent in Business Central
The Payables Agent focuses on the automation of vendor invoice processing – an area where finance teams commonly spend substantial time on data entry, matching, and validation.
What the Payables Agent Does
Invoice Capture from Email
The agent monitors a dedicated accounts-payable mailbox and identifies supplier invoices sent as PDF attachments. It automatically imports these documents into the “Inbound E-Documents” area for processing.
OCR and Data Extraction
Integrated OCR, powered by Azure Document Intelligence, extracts structured header and line-level data from vendor invoices, including vendor names, invoice numbers, amounts, dates, and line descriptions.
Vendor Identification and Safe Vendor Creation
The agent attempts to match invoice data with existing vendors. If no match is found, it asks an AP supervisor whether to create a new vendor. Any newly created vendor remains blocked until formally reviewed, supporting audit requirements.
Draft Purchase Invoice Creation
The agent uses historical patterns to suggest G/L accounts, item mappings, and dimensions. It provides explanations for each suggested field, improving transparency and user confidence.
Support for Three-Way Matching (Roadmap)
Future enhancements will allow the agent to automatically match invoices with purchase orders and receipts, improving compliance and reducing posting errors.
Human Review and Escalation
Where confidence levels are low, the agent surfaces tasks for reviewer action, ensuring quality control before invoices are approved or posted.
Overall, the Payables Agent reduces manual data entry, improves accuracy, accelerates invoice turnaround time, and free-up finance personnel’s time for higher-value analytical work.
Agent Playground: Building Custom Agents
Recognising that organisations have unique processes, Microsoft introduced the Agent Playground, a controlled environment within Business Central that allows power users and consultants to design, test, and refine their own autonomous agents.
Core Capabilities of the Agent Playground
Instruction-Driven Agent Design
Users can define what an agent should do using natural-language instructions – without writing AL code. This enables rapid prototyping and lowers the technical barrier for automation.
Interaction with Business Central UI
Agents built in the playground can navigate pages, read data, apply filters, click actions, and follow multi-step workflows, similar to how a trained human user would operate the system.
Permissions and Profiles
Each agent runs under a dedicated user identity with explicit permission sets, ensuring alignment with internal security controls and segregation of duties.
Troubleshooting and Transparency
The playground provides visibility into how the agent interprets instructions, the steps it performs, and any errors encountered. This feedback loop supports safe and iterative refinement.
The Agent Playground allows organisations to build industry-specific, department-specific, or workflow-specific automation that extends far beyond Microsoft’s out-of-the-box agents.
Business Impact of Agentic AI in Business Central
Improved Operational Efficiency
Agentic AI significantly reduces manual workload across sales, finance, and administrative functions. Tasks that previously required several steps from multiple personnel can now be executed end-to-end.
Higher Accuracy and Consistency
AI-driven interpretation ensures fewer data entry errors, more consistent calculations, and adherence to business rules.
Strengthened Governance
With every action logged and explainable, agentic AI supports compliance audits, internal controls, and policy enforcement.
Scalability without Additional Headcount
As transaction volumes grow, these agents operate continuously and predictably, offering a scalable approach to workload management.
Enhanced Customer and Vendor Engagement
By speeding up order acknowledgements and invoice processing, businesses improve service quality and partner relationships.
Conclusion
Agentic AI represents a major step forward for Business Central, enabling practical automation that mirrors the reasoning and behaviour of trained personnel while maintaining governance, oversight, and transparency. Organisations adopting these capabilities can expect measurable gains in productivity, accuracy, and operational agility.
As Microsoft continues to expand its agentic AI roadmap, businesses that invest early will be best positioned to scale efficiently and unlock new competitive advantages. Contact Us for a no-obligation demo.
