AI in ERP means that the system doesn’t just read data—it prepares the next action. Instead of someone opening a supplier invoice, comparing it to the purchase order, and posting it, the AI detects the discrepancy and submits the completed entry for approval.

The difference from traditional ERP functions: Rules only apply to cases that someone has defined in advance. AI can also handle situations that fall outside the norm.

The direction is clear: from supporting features toward deeply integrated, process-level agents. AI will be embedded more closely into existing workflows, helping reduce operational complexity over time. The goal is an ERP that doesn't just document, but actively contributes to efficiency.

Xentral is building AI capabilities into existing ERP processes step by step. The goal is to intelligently support operational workflows like order processing, purchasing, and accounting while reducing manual steps. The first applications are already available, with more being developed and tested. Businesses retain full control over decisions and approvals throughout.

The AI Sidekick can query data directly from the system. For quick lookups like open invoices, it returns live figures. For more complex queries, such as detailed reports on product groups or returns, the AI Copilot in Reporting is used to generate a structured report.

One important note: modifying data, such as updating selling prices, is not yet possible for security reasons. However, this is on the roadmap, with any such action requiring explicit user confirmation before it takes effect.

AI agents are designed to work through tickets in the background, pulling relevant data such as order numbers for returns or customer details for enquiries, then generating a suggested resolution. One of the first planned agents will also handle automatic order creation for orders that come in via email or the ticket system.