For the Mittelstand
AI for the Mittelstand: pragmatic, measurable, on site.
The German Mittelstand, the owner-managed and family businesses that carry most of the economy, does not need an AI transformation programme. It needs a handful of workflows that relieve people you cannot hire, run inside the ERP you already have, and pay for themselves in months. That is what we build, in German, with your team in the room.
What Mittelstand companies actually need
The question is rarely whether to use AI. It is how to get real relief for a stretched team without betting the company on a vendor or a consultancy.
If you are reading this in English: the Mittelstand is the German term for the country's small and mid-sized companies, typically owner-managed or family-owned, often world leaders in a narrow niche, with 50 to 2,000 employees, a lean IT department and an ERP system at the centre of everything. The culture is pragmatic, long-term and sceptical of fashions. We think that is exactly the right attitude toward AI.
The driver is usually not the technology but the skills shortage, the Fachkräftemangel. Order processing, service, accounting and sales support are short-staffed, the experienced people are approaching retirement, and open positions stay open for months. AI agents will not replace those people. They can take the repetitive part of the work off their desks so that the people you have go further, and so that knowledge does not walk out of the door.
We work incrementally: one workflow, one department head who wants it, measurable relief within weeks. We speak German, come on site, involve the works council early and hand over the knowledge so that you stay independent of us. Public funding programmes for digitalisation exist in Germany; we can point you to the relevant ones for your region and size, without promising what they will decide.
- Typical company50–2,000 employees, owner-managed or family-owned
- SystemsSAP, Microsoft Dynamics, proALPHA, abas, DATEV and similar
- Working modeGerman-speaking, on site across Germany, remote where sensible
- OutcomeWorking agents in your daily operations, and a team that can run them
Where Mittelstand companies usually stand
Open positions that stay open
The service department is missing two technicians, order processing runs on overtime, and the one person who knows the export documentation retires next year. Every process that depends on a single person is a risk, not just a cost.
An ERP at the centre and a lean IT team around it
Everything runs through SAP, Microsoft Dynamics, proALPHA, abas or a comparable system, plus DATEV for accounting and a shared drive full of spreadsheets. IT keeps it all running with a small team and has no capacity for experiments that create new systems to maintain.
Pressure to "do something with AI", scepticism about how
The advisory board asks, the trade association runs a seminar, a competitor shows a chatbot at the trade fair. At the same time, the owner has seen enough software projects overrun to want proof before commitment.
Data protection and the works council are part of the culture
Personal data of customers and employees, the GDPR and codetermination are not obstacles to be worked around. They are part of how the company operates, and an AI project that ignores them will not survive its first works council meeting.
Independence matters more than the newest tool
Nobody wants a second dependency on a vendor or a consultancy that has to be called every time something changes. Whatever is built must be understood, documented and maintainable by the company's own people.
What we do for the Mittelstand
Concrete, incremental, measured. Every engagement is sized so that one department can own it.
AI readiness assessment in two to three weeks
We interview leadership and department heads, look at your ERP, documents and data, and return a short written assessment: the three workflows worth automating first, what they need, and what to leave alone. Readable by the owner in an evening.
ii.Agentic workflows in the ERP you have
Order intake from emails and PDFs into the ERP, quote preparation, supplier communication, dunning and invoice matching, service ticket triage. Agents read, draft and prepare; your people check and release. See agentic workflows.
iii.Knowledge assistants that keep expertise in the company
Manuals, past service cases, drawings, contracts and the knowledge of retiring experts become a searchable assistant that answers with sources. New employees are productive sooner, and knowledge stops depending on one person.
iv.Training for management and departments
Half-day sessions for leadership on what AI can and cannot do, hands-on workshops for the teams who will use the agents, and the AI-literacy training the EU AI Act expects. Practical, in German, on your own examples.
v.Data protection, works council, EU AI Act
We design every workflow with your data protection officer and the works council in mind: what data is processed where, what stays in the EU, who checks what. Documented, so that the questions are answered before they are asked.
Operations and knowledge transfer
After launch we can run the agents for you as managed AI operations or train your IT and department to do it. Either way, the documentation and the know-how belong to you.
A typical path
Most Mittelstand engagements follow this sequence. Each step ends with a decision you can stop at.
On-site visit and intro
We visit the company, walk through the departments that hurt most and talk to the people doing the work. You get an honest view of whether AI helps here and what it would take.
Readiness assessment
Interviews, a look at ERP, documents and data quality, then a short written assessment with the first three use cases, an effort estimate and a note on data protection and works council involvement.
First pilot in production
One workflow, built inside your systems, measured against the baseline, with your department and IT involved. Typically six to eight weeks to a production-grade pilot that people use every day.
Expand, train, hand over
The next workflows follow the same pattern. In parallel we train your team, document everything and reduce our role until you run it yourselves, or we keep operating it for you if you prefer.
Frequently asked questions
We have a lean IT department. Can we even do this?
Yes, and that is the normal case. Most workflows connect to your ERP through existing interfaces and run on platforms your IT does not have to build. The IT effort in a pilot is typically a few days for access and review. We write the documentation so that IT can maintain the result.
Do we have to move our data to the cloud or to a US provider?
Not necessarily. Models can run on EU-hosted infrastructure, in a German data centre or, for some use cases, on your own hardware. We select the option that fits your data, your customers' requirements and your budget. Read more on sovereign AI and EU hosting.
How do you involve the works council?
Early and openly. We prepare a plain-language description of what the agent does, which data it processes and how employees are affected, and we join the meeting if that helps. In our experience the works council supports projects that relieve people and rejects projects it hears about last.
Are there subsidies for this?
Germany and the federal states run funding programmes for digitalisation and AI in small and mid-sized companies. Whether your company qualifies depends on size, region and the type of project. We can point you to the relevant programmes and help with the technical description, but we make no promises about approval.
Which processes pay off first in the Mittelstand?
Order intake and confirmation, quote preparation, service and maintenance knowledge, accounts payable and dunning, supplier communication, and internal helpdesks for HR and IT. See our pages on manufacturing and operations for typical examples.
Related pages
AI readiness assessment
A two- to three-week diagnostic across data, systems, skills, governance, processes and culture, ending in three first use cases and a 90-day plan.
Agentic workflow automation
Multi-step business processes automated end to end by LLM-powered agents across ERP, CRM, ticketing and email, with human approval steps built in.
Manufacturing & engineering
Agents for quoting, order intake, service knowledge, spare parts and quality documentation, built for ERP and PLM landscapes, OT/IT separation and export-controlled drawings.
AI training & workshops
Executive briefings, AI literacy for every employee, department workshops that end with working prompts, and agent-builder enablement for your power users. On site or remote, in German and English.
Let's find the first workflow worth automating.
A 30-minute intro call, no slides and no obligation. We listen, ask about your processes, and tell you honestly where AI agents would pay off and where they would not.