AI Agents on your AFAS data

AFAS is the core of personnel administration for many organizations. Through the connection between Conversed.ai and the AFAS HR-API, our AI Agents retrieve personnel data in real-time and securely, such as name, age, position, employment status, contract type, and leave balance. The agent then combines this data with employment conditions and HR documentation from systems like SharePoint. This way, an employee asking a question like “how many vacation days do I have left?” or “which policy applies to my contract?” receives a personal and correct answer immediately, without HR personnel needing to intervene.

In addition to HR, AFAS offers a full-fledged Finance API. Through GetConnectors and OData connections, our AI Agents retrieve virtually all financial data: financial transactions, booking days, general ledger accounts, and journal entries, as well as sales and purchase invoices with associated status and relationship information. For large datasets, post-calculations, revenue models, and open items.

One of the strong points of the AFAS integration is the degree of control. You define which GetConnectors are available, and thus exactly which fields an AI Agent can see: amounts, dates, relationships, cost carriers, cost centers, VAT codes, and more. This allows you to determine at the field level what an agent can and cannot retrieve. An HR agent will not have access to general ledger entries, and a finance agent will not see personal employee details. This granular authorization ensures that data is exclusively accessed where it is functionally and legally justifiable.

The AFAS integration goes beyond structured fields. Through AFAS's REST endpoints, FileConnector, ReportConnector, and ImageConnector, our AI Agents also retrieve underlying files: PDF invoices, reports, and images linked to a booking or relationship. In combination with Scan & Recognize, which enables AFAS to automatically read and book purchase invoices, the agent can perform analyses, conduct checks, or provide explanations based on the invoice data and mutations. This way, your team not only gets a number or amount but the complete document at their fingertips.

Personnel and financial data are among the most sensitive data within an organization. Therefore, the AFAS integration is built on the principle of least privilege: the agent only gets access to the fields necessary for the task. All personal data processed by our AI Agents is pseudonymized, meaning names, social security numbers, and other identifiable data are never directly visible. The processing is fully GDPR-compliant and aligns with the requirements of the European AI Act. This allows you to harness the power of your AFAS data without compromising security or compliance.

True added value is created when AFAS data is combined with knowledge found elsewhere in your organization. Through Retrieval-Augmented Generation (RAG), Conversed.ai links structured data from AFAS with documentation from systems like Zendesk, Freshdesk, SharePoint, and Google Drive. This allows an HR agent to enrich a personal leave balance from AFAS with the correct leave policy from your employee handbook; a finance agent can connect an invoice status with the corresponding purchasing policy. The result is an answer that is factually correct and contextually complete.

With Conversed.ai’s AI Optimization Studio, you maintain full control over what your AI Agents do with the AFAS data. You can see in real time which questions are being asked, how the agent is performing, and where there is room for improvement. Human feedback continuously improves the accuracy of the responses, while dashboards provide insights into frequently asked topics and trends within HR and Finance. This ensures that the integration remains not only secure and up-to-date, but also continuously delivers valuable insights for your broader business operations.