Integrations
AI can only run
what it
can see.
In most companies the data sits in five systems that do not talk to each other: the ERP knows the order, the TMS knows the tour, the payment provider knows the money — and in between, people retype. I connect those systems. Cleanly, monitored, and in a way an AI can build on.
Systems I connect
ERP
SAP
Deliveries, stock, advice notes and master data out of SAP — via IDoc, BAPI, OData or a plain file export.
SAP integration →Freight forwarding
CarLo
Orders, tours and shipment status out of the CarLo transport management system — into warehouse, customer portal and reporting.
CarLo integration →Payments
Stripe
The payment arrives — invoice, warehouse and accounting know immediately. Fees and payouts included.
Stripe integration →Payments
PayPal
PayPal transactions matched to the order and booked automatically — instead of a CSV export at month end.
PayPal integration →Your system is not on the list?
It does not matter. Give me the API documentation and I will connect it — including systems I am meeting for the first time. Interfaces follow a handful of patterns: authenticate, fetch, hand over, handle the failure. Once you have run those patterns in live operations, a new API takes days to read into, not months. ERP, inventory, freight and warehouse software, accounting, shops, payment providers, telematics, government portals — name the system and I will tell you which route it offers.
No API? Still connectable.
01
API & webhooks
REST, OData, SOAP — the direct route when it exists. Real time, both directions.
02
EDI & file exchange
IDoc, EDIFACT, XML, CSV over SFTP: the proven standard between ERP, forwarder and retail.
03
Email as an interface
The order arrives as a mail with a PDF; AI reads it and opens the case — structured, checkable, not retyped.
04
Excel & reports
The daily export someone pulls anyway becomes an automatic data stream, with plausibility checks.
05
Documents & photos
Delivery note, invoice, waybill: document AI turns a phone photo into a clean record.
06
Databases
MySQL, SQL Server, BigQuery, Google Sheets — if the data already sits somewhere, I collect it there.
Plenty of systems running every day in mid-sized companies have no modern API — or one that only opens for an extra licence fee. That is not a reason to keep retyping.
How I go about it
01
Read what the system offers
Which data lives where, who needs it when, and which route the system genuinely provides. The output is a one-page interface sheet, not a concept paper.
02
Prototype on real data
Within days the first real record travels from A to B. Only then do the actual problems show up: formats, edge cases, permissions.
03
Run it properly
On monitored servers, with retries, a log and explicit failure handling. Not a script on a forgotten office PC.
04
Let AI watch it
A delivery that does not arrive, a format that changes, a total that drifts — the integration reports it before accounting or the warehouse notices.
Classic interface projects end at hand-over. The data flows, the project is closed — until a field changes three months later and nobody notices. I build every connection to watch itself, because the point is not the interface. The point is a company where AI does not just see the data but runs the work.
Questions I get asked
You have never connected our system — is that a problem?
No. What matters is not whether I have seen a system before, but whether it offers a documented route: API, EDI, export or email. With the documentation on the table, the first prototype on real data is usually running within days.
What does an integration cost?
It depends on the route the system offers and on the number of edge cases — not on the size of the vendor. After the analysis I name a fixed frame before any cost arises.
Do we have to change our existing software?
As a rule, no. I use the routes your system already provides — exports, standard interfaces, mailboxes. Your ERP or TMS stays as it is.
What happens when the vendor changes its format?
That is what the monitoring is for: deviations in structure, volume or totals are detected and reported before wrong data travels on. The fix is then a matter of hours, not weeks.
Next step
Which two systems
should start
talking?
Tell me in two sentences what gets retyped today. I will tell you which route your system offers.
SAP is a trademark of SAP SE. CarLo is a trademark of Soloplan GmbH. Stripe is a trademark of Stripe, Inc. PayPal is a trademark of PayPal, Inc. I work independently and have no partnership or affiliation with the companies named. Names are used only to describe the technical connection. · Auf Deutsch: Integrationen bei Monster Intelligenz