We don’t digitalise your old problems.
We solve them.

It depends heavily on scope. A quick win – a clearly defined, recurring task – is often in production within a week and still removes noticeable manual work. Where several departments and systems are involved, four to eight weeks is more realistic. The biggest factor is usually not the technology but how quickly people on your side answer questions and provide access.

In four phases: analysis, concept, delivery, operation. After the analysis you receive an effort estimate with clear priorities before anything is built, so you decide on solid numbers rather than on assumptions. Throughout, you talk to the people who actually build your solution, not to a ticket system.

If you want it to, entirely. We run automation and AI processing on your own hardware, so documents and master data never leave the company. Incoming invoices, for example, are classified by local language models against your own supplier and cost-centre data — in-house and without any cloud connection.

No. We connect the systems you already have rather than replacing them: ERP, WMS, archive and industry-specific software stay where they are. We work at the interfaces – wherever someone currently moves data from one system into another. Replacing a system can make sense, but it is a separate decision and not a precondition for automation.

EDIFACT is a United Nations standard for exchanging business documents electronically between companies. Orders (ORDERS), despatch advices (DESADV) and invoices (INVOIC) are broken down into standardised segments that both sides can read automatically. In retail, manufacturing and logistics, EDIFACT is often a precondition for becoming a supplier at all.

Usually three things: a format (most often EDIFACT, sometimes XML or CSV), a transport channel (AS2, OFTP2, SFTP or a platform such as Cargoclix or Transporeon), and a mapping that translates the fields in your systems into the message. We deliver all three parts and connect them to your ERP or WMS, so you never have to run an EDI system yourself.

The nature of the data decides, not the technology. For documents containing personal data, prices or design information we recommend local models on your own hardware; for uncritical tasks a cloud model is often faster to introduce and cheaper. For document recognition, classification and data extraction, locally hosted models are generally sufficient today. With a cloud solution, operation naturally depends on the provider; an on-premise installation keeps running in your own house regardless.

When it replaces recurring, rule-based work, our projects usually pay for themselves in under a year – for quick wins considerably faster. Automation is not worth it for processes that change constantly, occur rarely, or are in truth an organisational problem: then changing the process is cheaper than automating it. When we reach that conclusion, we say so and advise against it.