Microsoft’s latest Business Central updates integrate AI-driven agents directly into the ERP platform, enabling organisations to define, test, and deploy natural language assistants that streamline routine business tasks while emphasising governance and transparency.
Artificial intelligence is being embedded ever deeper into enterprise resource planning, and Microsoft’s latest Business Central updates put agent authoring directly inside the ERP rather than as an ext...
Continue Reading This Article
Enjoy this article as well as all of our content, including reports, news, tips and more.
By registering or signing into your SRM Today account, you agree to SRM Today's Terms of Use and consent to the processing of your personal information as described in our Privacy Policy.
ernal bolt-on. According to Microsoft’s release notes and product guidance, the platform now offers a built-in Agent Designer and supporting toolkit that let organisations define, test and run AI-driven assistants that interpret natural language, invoke Business Central APIs and execute business tasks under governed permissions.
Setting up an agent begins with platform-level enablement and appropriate isolation for experimentation. Microsoft’s documentation and the AI Development Toolkit advise enabling the Custom Agent capability in a sandbox tenant and assigning roles such as AGENT-ADMIN and AGENT-DIAGNOSTICS so that development and diagnostics are limited to authorised users. Agents are created through a guided wizard where administrators specify the agent’s purpose, execution profile and the permission set it will run under; agents are treated like users and therefore must have explicitly defined access boundaries.
The functional core of each agent is a plain-language instruction set that describes intent, tasks and expected outputs. An example instruction used in early demonstrations reads: “You are a Business Central agent. When invoked, check all overdue receivables and create a work list of customers where the balance exceeds credit terms.” According to Microsoft’s guidance, the Agent Designer translates such instructions into orchestrated actions across pages, APIs and built-in business logic while logging activities for audit and review.
Profiles and permissions form the principal governance mechanism. Execution profiles determine which UI elements and actions an agent can reach, constraining read and write operations to minimise risk. Microsoft emphasises transparency and control; every action taken by an agent is recorded and can be reviewed, and many deployments favour a human-in-the-loop model where agents draft transactions or suggestions that a user must approve before final posting.
Practical use cases already showcased include sales order agents that monitor an inbox, extract customer intent, check availability and prepare quotes or orders while notifying a human reviewer, and payables agents that parse vendor invoices and generate draft postings for validation. Microsoft’s product blog and related posts position these agents as a way to reduce repetitive work and accelerate routine processing without extensive bespoke integration.
Outside Microsoft’s native tooling, an ecosystem of low-code and no-code vendors is emerging to simplify agent creation and connectivity. Third-party offerings such as Simple Agent Designer, Agenthost and other integrators claim one-click connectors to Business Central APIs, prebuilt templates and analytics dashboards to track agent performance and business outcomes. Industry commentary highlights that such tools can lower the bar for non-developers to build operational agents, but also stresses the need to align third-party connectors with an organisation’s security and compliance posture.
Generative AI and copilot frameworks are being positioned as the standardised protocol for connecting agents to knowledge sources and ERP functions. Microsoft’s wider Dynamics 365 roadmap and Copilot Studio aim to provide consistent patterns for agents to access data and services, which vendors say should improve reliability and scalability across deployments. At the same time, early adopters are urged to treat agents as governed automations: define clear intent, test extensively in sandboxes, limit privileges by profile and retain human oversight for material financial or customer-facing actions.
The introduction of first-class agent capabilities into Business Central signals a shift from rule-based workflows to intent-driven automation embedded inside core systems. For organisations weighing adoption, the immediate benefits are operational efficiency and reduced manual entry; the longer-term challenge will be establishing robust governance, auditing and change controls so that autonomous actions remain aligned with enterprise risk and compliance requirements.
Source: Noah Wire Services
