How to Build Custom AI Agents in Dynamics 365 Business Central ( Update)
Most ERP systems require manual oversight for routine tasks like inventory replenishment and sales order validation. Microsoft's new Agentic AI functionality in Business Central lets you create automated workflow agents that handle these tasks - no coding required. Here's how to build your first custom agent in under 10 minutes.
What Are Business Central AI Agents?
ERP systems like Dynamics 365 Business Central contain dozens of routine processes that follow predictable patterns - validating sales orders, checking inventory levels, reconciling accounts. Until now, these required manual attention or complex custom development. Microsoft's February 2026 update introduces native AI agents that automate these workflows.
Unlike traditional automation that requires fixed rules, AI agents in Business Central can interpret instructions and make context-aware decisions. The system provides pre-built agents for common tasks, but the real power lies in creating custom agents tailored to your specific business needs.
Key difference: While Microsoft's previous AI features focused on assisting users, these agents operate autonomously. They can complete entire workflows from start to finish, logging each action in Business Central for audit purposes.
Sales Validation Agent Demo
Microsoft includes several template agents with the update, including a sales validation agent. This agent automatically reviews sales orders for common issues like incorrect pricing, missing customer information, or inventory shortages.
At the 2:15 mark in the video, we see the agent in action. After activation, it processes all open sales orders, creates a validation report, and presents its findings for review. The entire workflow follows the instructions defined in the agent's configuration - no human intervention required until the final approval step.
Implementation insight: The sales validation agent demonstrates how Business Central agents combine AI decision-making with ERP system actions. The AI interprets the instructions, examines the data, then uses standard Business Central APIs to execute changes - all while maintaining a complete audit trail.
Building an Inventory Replenishment Agent
Inventory management is one of the most time-consuming aspects of ERP systems. The demo shows how to create a custom agent that automatically replenishes stock when levels fall below zero.
Starting from scratch at 4:30 in the video, we see the agent creation process:
- Define the agent's purpose (monitor and replenish inventory)
- Set appropriate permissions (item and requisition access)
- Write clear instructions for the AI to follow
The key advantage is that this agent doesn't just alert about low stock - it proactively creates the requisition worksheet lines needed to correct the imbalance. This eliminates the manual work of checking each item's availability and calculating order quantities.
Agent Instructions Structure
Effective agents require well-structured instructions. At 6:45, the video reveals the template Microsoft recommends:
- Introduction: Define the agent's role and scope ("You are an inventory management agent...")
- Process steps: Detailed actions to perform ("Check projected available balance for all items...")
- Decision logic: Conditions and responses ("If quantity is negative, create a requisition line...")
- Output requirements: What to record or report ("Log all actions in the agent history...")
This structure ensures the AI understands both what to do and why. The inventory agent's instructions run about 15 lines - enough detail to handle the workflow without being overly prescriptive.
Running Your Custom Agent
Once configured, running an agent takes one click. At 8:20, we see the inventory agent process all items in seconds:
- Agent scans item availability records
- Identifies conference tables with negative projected balance
- Creates corresponding requisition worksheet lines
- Logs all actions in Business Central's history
The demo uses a sandbox environment, which Microsoft recommends for testing agents. This lets you verify an agent's behavior before deploying to production. The completed requisition worksheet serves as both the agent's output and your starting point for the purchasing process.
Copilot Billing Setup
AI agents consume Copilot credits, requiring proper Azure billing setup. The video shows the three-step process at 12:30:
- Create an Azure environment (if none exists)
- Establish a billing plan in Power Platform Admin Center
- Link the Azure environment to your Business Central instance
Microsoft provides limited free credits - about 4 agent runs in the demo. For production use, you'll need to monitor usage and adjust your billing plan accordingly. The setup is straightforward but requires administrator privileges in both Azure and Business Central.
Current Limitations
While powerful, the initial agent release has some constraints shown in the demo:
- No scheduling: Agents can't automatically run hourly/daily (must trigger manually)
- Credit consumption: Complex agents may exhaust credits quickly
- Sandbox recommendation: Microsoft advises testing in non-production environments
Future updates will likely address these limitations. The inventory agent demonstrates the potential - even in its current form, it can save hours per week on routine ERP maintenance tasks.
Watch the Full Tutorial
See the complete walkthrough of creating and running custom AI agents in Business Central. The video demonstrates both Microsoft's template agents and building a custom inventory replenishment agent from scratch.
Key Takeaways
Microsoft's new Agentic AI functionality brings powerful automation to Dynamics 365 Business Central without requiring coding skills or external tools. While the initial release has some limitations, the ability to create custom workflow agents represents a major leap forward in ERP automation.
In summary: Business Central agents let you automate routine tasks by writing natural language instructions instead of code. Start with template agents, then build custom solutions like the inventory replenishment agent shown in the demo. Always test in sandbox environments and monitor your Copilot billing usage.
Frequently Asked Questions
Common questions about Business Central AI agents
AI agents in Business Central are automated workflows that can perform specific tasks within the ERP system. Microsoft's February 2026 update introduced the ability to create custom agents without needing external tools like Copilot Studio.
These agents follow predefined instructions to automate processes like sales order validation or inventory replenishment. Unlike traditional automation that requires fixed rules, they can interpret context and make appropriate decisions within their defined scope.
- Operate autonomously to complete entire workflows
- Maintain detailed logs of all actions taken
- Can be created through Business Central's native interface
To create a custom agent, navigate to the Agents section in Business Central and select Create. You'll need to define three key components: the agent's purpose, appropriate permissions, and clear operating instructions.
The system provides templates to help structure your agent's workflow, like the sales validation template shown in the demo. For the inventory replenishment agent, we defined specific criteria for when to create purchase requisitions based on stock levels.
- Start with the agent creation wizard in Business Central
- Define permissions matching the agent's responsibilities
- Use natural language instructions the AI can interpret
The demo showed an agent that checks item availability in Business Central. When it finds negative projected available balances (indicating potential stockouts), it automatically creates requisition worksheet lines to replenish inventory.
This basic example demonstrates how agents can handle routine ERP maintenance tasks without human intervention. The agent examined each item's availability, applied the predefined replenishment logic, and generated the necessary procurement documentation.
- Scans all inventory items for negative balances
- Creates corresponding purchase requisitions
- Logs all actions in Business Central's history
No coding is required to create basic agents. The interface uses natural language instructions similar to how you'd explain the task to a human employee. You write the steps you want the agent to follow in plain English (or your preferred language).
However, understanding Business Central data structures and workflows helps create more effective agents. The system provides templates to guide your instruction writing, but familiarity with how your ERP processes work leads to better automation design.
- Uses natural language processing instead of code
- Microsoft provides instruction templates
- ERP knowledge improves agent effectiveness
Current limitations include the inability to schedule recurring agent runs (though this may change in future updates) and dependence on Copilot billing credits. The demo showed that about 4 agent runs consumed the free credits Microsoft provides.
Other constraints include the recommendation to test agents in sandbox environments first, and the need for proper Azure billing setup before production use. Microsoft is actively developing this functionality, so expect these limitations to evolve over time.
- No built-in scheduling for recurring runs
- Consumes Copilot credits that require billing setup
- Microsoft recommends sandbox testing first
Setting up billing requires creating an Azure environment and linking it to your Business Central instance through the Power Platform admin center. You'll need administrator access to both systems to complete the configuration.
The process involves creating a billing plan that covers the AI operations, then associating it with your Business Central environment. Microsoft provides some free credits to get started, but production use will require proper Azure billing configuration.
- Requires Azure environment setup
- Configured through Power Platform admin center
- Free credits available for initial testing
Microsoft provides pre-built agents like the sales order agent and payables agent that handle common ERP workflows. These offer standardized solutions for frequent business processes across industries.
Custom agents let you create specialized automations for your unique business requirements. The inventory replenishment agent in the demo is an example of a custom solution you couldn't achieve with the standard agents, tailored to specific inventory management needs.
- Built-in agents handle common cross-industry workflows
- Custom agents address unique business requirements
- Both types maintain complete action logs
GrowwStacks helps businesses implement AI automation in Dynamics 365 Business Central, from simple task agents to complex multi-step workflows. Our consultants have deep experience with both Business Central and AI automation technologies.
We can design custom agents for your specific ERP needs, set up the required Azure billing infrastructure, and train your team on agent maintenance and optimization. Our implementation process ensures your agents deliver measurable efficiency gains.
- Custom agent design for your unique workflows
- Azure billing setup and cost optimization
- Training and documentation for your team
Automate Your Business Central Workflows Today
Manual ERP tasks waste hours each week that could be spent growing your business. Our Dynamics 365 automation experts can implement custom AI agents tailored to your operations in as little as 2 weeks.