DYNEXAL RESEARCH • AI + BUSINESS CENTRAL

AI Is Changing Business Central Development: What Developers Should Prepare for in 2026

Artificial intelligence is moving from a separate productivity tool into the core of enterprise applications. Microsoft Dynamics 365 Business Central is part of this transition, with AI capabilities, agent scenarios and developer tooling becoming increasingly relevant to the platform.

The key question for developers: not simply “Will AI write AL code?”, but “How should a Business Central developer work when AI becomes part of both the development process and the ERP solution itself?”

1. Business Central is moving toward agentic ERP

Traditional ERP systems primarily record and process business transactions. The emerging model connects business data with intelligence, recommendations and controlled actions.

For Business Central developers, this means AI is increasingly connected to real business processes rather than being limited to a chatbot. Examples include sales activities, purchase workflows, expense processing, inventory decisions, financial analysis and developer productivity.

2. What does this mean for AL developers?

AI does not remove the need for Business Central development knowledge. Instead, the workflow can shift from requirement-to-code toward AI-assisted design, implementation, testing and review.

  1. Understand the business requirement.
  2. Use AI to explore solution options and generate a draft.
  3. Implement or refine the AL solution.
  4. Generate and run tests.
  5. Review security, performance and upgradeability.
  6. Validate the final behavior before deployment.

The developer remains responsible for business logic, architecture, data integrity, permissions and production behavior.

3. Coding agents are becoming relevant to AL development

Microsoft's current AL developer tooling includes AI-agent capabilities for development workflows. Depending on the development surface and environment, these tools can assist with tasks such as building projects, publishing extensions, working with symbols, diagnostics and debugging.

This changes the developer workflow from isolated code completion toward larger tasks such as planning, implementation, testing and troubleshooting. Human review remains essential.

4. AI + AL is a new development skill

Business Central extensions can use AI to provide specialized business experiences. A practical example is an AI Customer Analyzer that combines approved customer information with AI reasoning to produce a readable summary of sales trends, outstanding balances, overdue invoices and suggested follow-up actions.

The important architecture is:

Business Central data → secure context → AI processing → validation → user-facing recommendation → optional controlled action

5. AI agents could change ERP workflows

Consider a sales-order scenario. A traditional workflow requires a user to check the customer, products, availability and pricing before creating an order. An agent-assisted workflow can help interpret a request, retrieve approved information, prepare a transaction and present it for human approval.

The critical distinction is that an agent is not simply answering a question; it is participating in a business process. Permissions, validation, auditability and approval boundaries therefore become part of the solution design.

6. MCP is an important concept for BC developers

Model Context Protocol (MCP) provides a structured way for AI systems and agents to discover and use tools and business context. For Business Central, the important engineering question is how an AI agent can interact with ERP data and operations without bypassing application security.

7. What Business Central developers should learn now

Core BC: AL, tables, pages, reports, codeunits, events, interfaces, APIs, extensions, permissions, testing, performance and upgrades.

Integration: REST APIs, JSON, HttpClient, OAuth 2.0, webhooks, Azure services and Power Platform.

AI: generative AI fundamentals, prompting, structured output, grounding/RAG concepts, AI evaluation, agents, MCP and responsible AI.

Cloud: Azure, Microsoft Entra ID, Azure OpenAI concepts, telemetry and CI/CD.

8. The developer role is becoming more architectural

As AI-assisted development improves, typing code becomes only one part of the job. The harder questions are architectural: what should be built, where should logic live, what data should AI access, which actions should an agent perform, when is human approval required, and how should the solution be tested?

That makes Business Central domain knowledge, solution design and engineering judgment increasingly important.

9. A practical AI project for BC developers

Consider an AI-powered Vendor Analysis extension. Approved vendor ledger entries, purchase history, payment history and outstanding balances can be transformed into a controlled AI context. The result can explain trends and exceptions while keeping the final business decision with the user.

Production principle: the value is not the AI-generated sentence. The value is the complete architecture around data access, security, validation, user experience and controlled action.

10. Recommended learning path

  1. Become strong in AL and Business Central architecture.
  2. Master APIs and integrations.
  3. Learn generative AI and prompt engineering fundamentals.
  4. Explore Business Central AI and Copilot capabilities.
  5. Study the AL developer tooling for AI experiences and agents.
  6. Build a small AI extension with human review.
  7. Learn permissions, data governance and responsible AI.
  8. Explore agents and MCP after understanding APIs and security.

Conclusion

Business Central development is entering a new phase in which AL, AI, agents, APIs, cloud services and human governance work together. The strongest preparation is not to abandon AL, but to add AI to an existing Business Central engineering foundation.

AL Developer → AI-Assisted AL Developer → Business Central AI Developer → AI Agent Developer → AI + Business Central Solution Architect

Sources & references

← Back to Dynexal Insights