How an AI layer works in payroll.

The payroll in German companies has long been digitized. DATEV, Sage, and Lexware process millions of payrolls every month. Nevertheless, HR departments spend a significant portion of their working time on manual data entry. The paradox: The actual calculations run automatically, but the path of the data into the system remains manual labor.

An AI layer solves this problem. It acts as an intelligent layer between employees, HR systems, and payroll software. Instead of manually transmitting data, the AI layer interprets incoming information and forwards it in a structured manner.

The Reality: Proven Systems with Manual Effort

DATEV LODAS and payroll software dominate the German market. Many companies also rely on Sage or Lexware. These systems are mature, legally compliant, and continuously adapted to new legal requirements.

The challenge is not in the software itself. It lies in the process before it. Before DATEV can create a payroll, all relevant data must be in the system: working hours, sick leave notifications, vacation days, changes in bank details, or tax classes. This information reaches the HR department through various channels: by email, through HR portals, as scans, or via phone.

System of Record: The Foundation Remains

DATEV and comparable solutions are known as Systems of Record. They form the binding data basis for all payroll-relevant processes. Every booking, every calculation, every proof runs through this system. Legal requirements such as GoBD-compliant archiving are implemented here. Replacing these systems is not up for debate for most companies. An overview of the integration options for DATEV can be found at project b. in their DATEV documentation.

The Problem of Manual Data Entry

The bottleneck occurs during data entry. An employee informs by email that their bank details have changed. The HR department must manually transfer this information to DATEV. The same applies to sick leave notifications, changes in working hours, or new employees.

For larger companies, this effort adds up significantly. Each manual entry carries the risk of errors: transposed numbers in the IBAN, incorrectly transferred dates, forgotten changes. Correcting such errors takes additional time and can, in the worst case, lead to incorrect payrolls.

System of Engagement: The New Interaction Layer

This is where the concept of the System of Engagement comes into play. While the System of Record retains data control, the System of Engagement takes on interaction with users. It receives requests, processes them, and forwards the structured data to the core system. Gartner describes this architecture as a supplement to existing systems, not as a replacement.

An AI layer is such a System of Engagement. It understands natural language, recognizes contexts, and can handle various data sources. Fraunhofer IAO describes how AI systems enhance existing enterprise software rather than replace it.

How the AI Layer Works

The AI layer acts as a mediator. It receives data from different sources and translates it into the format needed by the payroll system. The process follows a clear pattern:

1. Data Entry: Information reaches the system, for example, via email, form, or app.

2. Interpretation: The AI layer analyzes the content and determines what type of change it is.

3. Validation: The system checks whether all necessary information is present and plausible.

4. Structuring: The data is brought into the required format.

5. Transmission: The AI layer forwards the structured data to DATEV, Sage, or the respective target system.

6. Confirmation: Employees and the HR department receive feedback on the processing status.

Practical Example: Change of Bank Details

An employee changes their bank. Previously, the process was as follows: write an email to the HR department, wait for processing, hope that no errors occur. With an AI layer, the procedure changes fundamentally.

The employee enters the new IBAN in an app or shares it via voice message. The AI layer automatically recognizes that this is a change in bank details. It validates the IBAN for formal correctness, assigns the change to the correct employee record, and transfers it to the payroll system. The HR department receives a notification for approval, and the employee receives a confirmation.

Benefits for Companies and Employees

Implementing an AI layer significantly reduces manual effort in the HR department. Routine tasks, such as transferring master data, are eliminated. The error rate decreases because fewer manual entries are made. Employees benefit from faster processing times and direct feedback. KPMG shows in a study how AI applications in human resources enable efficiency gains while increasing employee satisfaction.

At the same time, the proven payroll system remains unchanged. Companies do not need to replace DATEV or Sage. The AI layer connects to the existing infrastructure and enhances it with an intelligent input layer.

project b. as an AI Layer for Existing Payroll Systems

project b. is an example of such an AI layer in practice. The platform connects with DATEV, Sage, Lexware, and other payroll systems without changing their functionality. Employees record data via a central interface: changes to master data, working time records, sick leave notices, or expense reports. The integrated AI assistant RITA supports data entry, identifies inconsistencies, and answers follow-up questions directly in the system.

The data flow operates automatically. Entries are validated, translated into the respective target format, and handed over to the payroll system. For DATEV users, this means: The familiar software remains the leading system while project b. simplifies the entire process before it. Digital personnel files, pay slip distribution, and change logs are centrally available. Tax consulting firms that bill for multiple clients significantly reduce the effort for reconciliations.

Conclusion

An AI layer in payroll does not replace existing systems. It makes them more accessible. The actual computation remains with DATEV, Sage, or Lexware. The AI layer takes on the task of bringing data from the real world into these systems without humans having to act as translators.

For companies, this means: less routine work, fewer errors, faster processes. For employees: easier communication with the HR department. For IT: no replacement of proven systems, but a meaningful addition.

Sources

1. project b. – DATEV integration and interfaces

2. Gartner – Definition of System of Engagement

3. Fraunhofer IAO – AI enhances enterprise software

4. KPMG – Artificial Intelligence in HR

5. DATEV – Payroll Product Overview