Business Central 2026 Wave 1: The Rise of Autonomous ERP Agents in Version 28

2. Autonomous Agent Architecture & ReAct Loops

Unlike traditional batch background jobs (`Job Queue Entries` in AL) that follow static logic trees, Autonomous Agents leverage ReAct (Reasoning + Acting) Execution Loops. When an agent processes a document, it completes a multi-stage cognitive cycle:

1. Observation: The agent detects an incoming document payload (e.g., an e-invoice attached to an email or a photo uploaded to the mobile app).
2. Reasoning: The agent evaluates vendor history, payment terms, tolerance limits, and historical G/L account distribution patterns stored in Dataverse.
3. Action Planning: The agent constructs a sequence of native AL web API calls or Codeunit invocations to create draft documents.
4. Execution & Telemetry: The agent executes the transactions within an isolated sandboxed session, logging telemetry to Azure Application Insights.

3. Data Model & Telemetry Schema in AL

Autonomous Agents require strict schema governance to prevent runaway transactions. In Business Central v28, Microsoft introduces new system tables for Agent Policy, Tolerance Sets, and Execution Audit Trails:

4. Step-by-Step Setup & Governance Matrix

Configuring autonomous agents in a production Business Central environment requires a 4-tier security checklist cross-referenced with official Microsoft Learn documentation (`learn.microsoft.com`):

5. Production AL Code: Custom Agent Action Handler

AL developers can register custom Business Central functions as callable tools for Autonomous Agents. Below is a production-ready AL Event Subscriber demonstrating capability registration and tolerance enforcement:

codeunit 50150 "Autonomous Expense Agent Action"
{
    Access = Public;
    Subtype = Normal;

    [EventSubscriber(ObjectType::Codeunit, Codeunit::"Copilot Capability", 'OnRegisterCopilotCapability', '', false, false)]
    local procedure RegisterCapability()
    var
        CopilotCapability: Codeunit "Copilot Capability";
    begin
        // Register custom AI Agent action within Business Central v28 runtime
        CopilotCapability.RegisterAccountantAgentAction(
            'DATADAUR_EXPENSE_AGENT',
            'Automated Corporate Expense Matching & Outlier Audit',
            Database::"Expense Header"
        );
    end;

    [EventSubscriber(ObjectType::Table, Database::"Expense Header", 'OnBeforePostExpenseHeader', '', false, false)]
    local procedure ValidateAgentTolerance(var ExpenseHeader: Record "Expense Header"; var IsHandled: Boolean)
    var
        AgentSetup: Record "Agent Tolerance Setup";
        ApprovalMgt: Codeunit "Approvals Mgmt.";
    begin
        if IsHandled then
            exit;

        AgentSetup.Get();
        
        // Enforce strict Human-in-the-Loop thresholds for high-value transactions
        if ExpenseHeader."Total Amount" > AgentSetup."Auto-Approve Threshold" then begin
            ExpenseHeader."Agent Processing Status" := ExpenseHeader."Agent Processing Status"::"Pending Human Approval";
            ExpenseHeader.Modify(true);
            
            ApprovalMgt.CreateApprovalRequests(ExpenseHeader.RecordId);
            Message('Expense %1 exceeds auto-approval threshold of %2. Escalate to Finance Manager.', 
                ExpenseHeader."No.", AgentSetup."Auto-Approve Threshold");
                
            IsHandled := true;
        end else begin
            ExpenseHeader."Agent Processing Status" := ExpenseHeader."Agent Processing Status"::"Auto-Approved by Agent";
            ExpenseHeader."Approved Date" := CurrentDateTime;
            ExpenseHeader.Modify(true);
        end;
    end;
}

6. Edge Cases, Rate Limits & Failover Protocols

Deploying AI agents into high-volume financial environments requires robust edge-case failover handlers:

• API Rate Limit Throttling: If Azure OpenAI endpoint limits are reached during batch invoice ingestion, the agent gracefully suspends execution and schedules a retry in Job Queue Entry.
• Currency Rounding Variances: Multi-currency vendor invoices with fractional cent rounding discrepancies trigger a automatic Rounding Deviation buffer G/L account write-off up to $0.05.
• Unrecognized G/L Account Mappings: If an expense description lacks historical G/L account correlation, the agent routes the item to a Pending Mapping Review queue instead of posting to misclassified accounts.

7. Measurable Business Impact & Key Metrics

Enterprise organizations implementing Autonomous ERP Agents in Business Central v28 achieve transformational operational gains:

Autonomous ERP Agents in Microsoft Dynamics 365 Business Central 2026 Release Wave 1 represent the most significant architectural evolution since the cloud migration from Dynamics NAV. By combining ReAct cognitive loops with strict AL governance, enterprise finance teams eliminate transactional overhead while retaining full oversight.