Connect AFAS Profit to Claude, ChatGPT or any MCP client. Ask about employees, absence, invoices, projects and hours in plain language, across years of data and the other tools you use.
The company average was 3.8%. Two departments were above 5%:
| Department | Last quarter | Same quarter last year |
|---|---|---|
| Logistics | 6.4% | 5.1% |
| Customer Service | 5.7% | 4.9% |
Four in Logistics and two in Customer Service.
AFAS Profit runs your HR, payroll, finance and projects. Answering a quick question still means reports, filters and Excel.
"Show me sick leave by department this quarter" turns into a report, an export and a pivot table.
Your bank, webshop and other tools sit outside AFAS. Real questions need them side by side.
AFAS shares data through GetConnectors, one per call. Questions across modules and years mean many calls.
A do-it-yourself MCP server means REST code, token handling and hosting.
Peliqan syncs your AFAS GetConnectors into its built-in data warehouse and gives your AI one MCP server.
Ask about employees, absence, payroll cost, invoices, debtors, projects and hours. Your AI writes the query and gives you the answer.
Each GetConnector becomes a table in Peliqan's data warehouse, so one question can combine HR, finance and projects.
Join AFAS with HubSpot, Exact Online, your bank file or any other connected app in one question.
Peliqan holds the AFAS token, so your AI app and the AI vendor never see it.
Peliqan reads every GetConnector you enable in AFAS Profit, so your AI sees exactly the data you choose. Typical questions per area:
| Area | Examples of what you can ask |
|---|---|
| Employees and contracts | Headcount per department, contracts ending soon, changes since the start of the year |
| Absence and leave | Sick leave rate per department, leave days per employee, leave requests |
| Payroll | Payroll cost per department against last year, overtime per department |
| Finance | Sales invoices, open debtor and creditor invoices, receivables aging |
| Projects and hours | Hours per project, billable against non-billable hours, hours and costs against budget |
Four example questions. Each one is a single prompt.
Most questions run as SQL on the copy in Peliqan's data warehouse, so big questions across modules and years don't load AFAS.
You set up an app connector in AFAS once and enter its credentials in Peliqan. Your AI app never sees them.
You set the sync schedule per connection, such as hourly or daily. Answers reflect the last sync.
Nothing to install, and no code to write.
In AFAS Profit, enable the GetConnectors you need on an app connector and enter its credentials in Peliqan. See the AFAS setup guide.
Add https://mcp.peliqan.io/mcp as a custom connector in Claude, ChatGPT or Gemini, then sign in to Peliqan.
For example: "Give me the current headcount per department and how it changed since the start of the year."
As of October 2026, AFAS offers no MCP server itself. Some third parties do, or you can build your own.
| Peliqan MCP | Build your own | |
|---|---|---|
| How it works | Synced copy in a data warehouse | Calls GetConnectors live |
| Questions across modules and years | Yes, as SQL across all your GetConnectors | One GetConnector per call |
| Other apps in one question | 300+ in the same server | What you build |
| Where the token lives | With Peliqan, never in your AI app | In your own config |
| Setup | Enable GetConnectors, add a token, add one URL | REST code plus hosting |
| Pricing | €15 per connection per month, 7-day free trial | Your developer time |
MCP is one part. The same connection also feeds reporting, transformations and syncs.