AI in Business Central: from rules to Copilot and agents

AI in Business Central does feel like it appeared overnight. One moment artificial intelligence was something discussed in theory; the next it was embedded in many of the workflows and processes that we use every day.

In reality, artificial intelligence has been part of Business Central for much longer than many of us realise. What has changed is its visibility, accessibility and how much responsibility organisations are now giving it. From long‑established rules‑based automation, through prediction and forecasting, to Copilot and AI‑driven agents, the AI in Business Central evolution has been gradual rather than sudden.

It is, however, important to understand how this progression has happened. When one person talks about AI in Business Central, they may be referred to something completely different to another one.

What does “AI in Business Central” actually mean?

When people search for AI in Business Central, they are rarely talking about a single feature. Some mean automation that has existed for years while others mean predictive tools such as forecasting. Increasingly, it is Copilot, natural language interaction and autonomous agents that users mean when they refer to AI in Business Central.

Rules-based automation: the foundations of ERP intelligence

Long before AI became a buzzword, ERP systems were already removing manual effort through rules‑based automation.

This familiar “if this happens, then do that” logic underpins much of Business Central today. One example is stock planning. If inventory drops below a defined level, the system recommends replenishment based on predefined rules.

These processes are predictable and explainable. Given the same inputs, they will always behave in the same way. It is this reliability is that makes rules‑based automation the backbone of core ERP workflows such as Order-to-Cash and Procure-to-Pay.

A major downside of this automation is its lack of flexibility. When scenarios fall outside the rules that have been defined, these systems cannot adapt without human intervention.

Machine learning in Business Central: prediction before conversation

Machine learning added a new layer of intelligence by introducing prediction. Instead of users defining every rule, algorithms learn patterns from historical data and use those patterns to anticipate future outcomes. Business Central has used this approach for years through capabilities such as sales forecasting and late payment prediction.

These capabilities help teams plan ahead and identify risks. However, it is still a passive form of artificial intelligence. Machine learning can highlight what might happen, but it does not explain why in human terms, nor does it act on those insights by itself.

Machine learning remains essential for forecasting and optimisation, but it was never intended to be the final step.

Generative AI: why Copilot feels different

Generative AI is where the experience changes most noticeably for users. Instead of working only with structured data and predefined outputs, generative AI introduces natural language interaction and context-aware responses. Users can ask questions, request summaries and generate content using everyday language, rather than navigating reports or dashboards.

In Business Central, this is where Copilot sits. It allows users to interact with ERP data conversationally and to generate explanations or summaries instead of simply retrieving figures.

Generative AI does not replace rules‑based automation or machine learning. It sits on top of them, making existing intelligence easier to access and apply.

From Copilot to agents: what agentic ERP means

AI agents are the next stage of the evolution. They are designed to do more than respond to prompts. They can work through parts of a process end-to-end within defined workflows and approval controls. An agent might read an incoming email, interpret intent, retrieve data from Business Central, create or update records, and review the outcome before completing the process.

This is a shift from assistance to execution and is very much the direction of travel for AI in Business Central. The introduction of sales agents, payables agents, expense agents and tools for building custom agents makes it clear that Microsoft does not plan to remove users from processes altogether. Instead, they are directing human effort away from repetitive execution and towards oversight, judgement and exception handling.

What this means in practice