Payroll Engine integrates through three channels: pre-built adapters, AI agents over MCP, and the open REST API. Adapters are the turnkey path — inbound connectors import HR master data from your existing systems, and outbound connectors post the computed results back to your ERP and finance. Because results flow back into the systems you already run, Payroll Engine slots into your landscape instead of demanding a rip-and-replace — and the adapter channel lets you start at any of three commitment levels.
Payroll Engine connects through three channels. Pre-built adapters are the turnkey path; AI agents connect through the Model Context Protocol (MCP); and the same open REST API powers everything underneath. Choose the channel that fits your stack — they all drive the identical calculation core.
Ready-made bidirectional connectors for common HR, ERP, and EOR systems — Personio, Silae, AFAS, Dynamics 365, SAP SuccessFactors, Sage, Papaya Global, PayFit, Odoo, and CSV. Inbound import of master data, outbound export of computed results, with secret handling and transient retry built in.
The Model Context Protocol exposes payroll as structured, permission-controlled tools for AI agents — not a chatbot, a tool interface. The open-source MCP Server gives read-only access to query employees, cases, and results. MCP Server Pro adds write and consolidation tools, so agents can enter case data, trigger payruns, and run cross-country queries.
The engine is an ASP.NET Core REST API: every capability — tenants, cases, payruns, results, and reports — is available programmatically. Query with OData, integrate with the typed .NET Client SDK, and embed payroll directly into your own platform. This is the foundation that adapters and MCP are built on.
Within the adapter channel, the same bidirectional framework supports an adoption ladder — from a zero-risk parallel run to a full payroll-engine deployment. Each scenario adds commitment while reusing the identical integration plumbing, so progressing from one to the next requires no re-integration.
Run Payroll Engine as a parallel (shadow) payroll next to your current production system — without switching anything. The lowest-risk way to prove compliance depth before any decision.
Keep your current payroll for production and add the forward-looking analysis it lacks — what-if, budgeting, and cost projection — computed by the same engine that runs the real payslip.
See Payroll Forecasts for the two forecast operating models in detail.
Introduce Payroll Engine as your new production payroll inside your existing HR/ERP landscape. Payroll Engine replaces only the calculation core — your HR master, ERP, and finance systems stay in place.
| Scenario | Adapter | Payrun type | Existing system | License | Typical use |
|---|---|---|---|---|---|
| 1 · Validate an existing payroll | Inbound | Preview | Stays in production | Forecast | Evaluation, audit, compliance check |
| 2 · Extend with forecasts | Inbound (+ optional export) | Forecast | Stays in production | Forecast | What-if, budgeting, cost projection |
| 3 · Extend your HR/ERP | Inbound + Outbound | Legal | HR/ERP retained, payroll core replaced | Full | Production payroll, multi-country rollout |
The Model Context Protocol (MCP) exposes payroll as structured, permission-controlled tools for AI agents — not a chatbot, a tool interface. It comes in two tiers: read-only query access, and write access for full agent-driven workflows.
The open-source MCP Server gives agents read access to query employees, cases, payruns, and results through a defined interface — without ever changing payroll state.
MCP Server Pro adds write and consolidation tools, so agents can enter case data, trigger payruns, and run cross-country consolidation queries — every action permission-scoped and audited.
Every capability of the engine is exposed as an ASP.NET Core REST API — tenants, cases, payruns, results, and reports. Query with OData, use the typed .NET Client SDK, and embed payroll directly into your own platform. The API is part of the open-source Payroll Engine, so there is no proprietary integration layer to license.