01
No in-house AI team
IT keeps operations running. Specialist teams know the process. In between, someone who connects both and delivers is missing.
AI for mid-sized companies
AI for mid-sized companies Automate processes and make knowledge usable
I build custom AI solutions for concrete business processes, internal knowledge and Microsoft 365 environments. First I check process, data risk and value. Then a solution that fits your landscape and that your team can keep running. Remote, documented, without an unnecessary system swap.
The problem
AI licences are there. Day-to-day value often is not.
Copilot or a chatbot subscription do not replace process work. Mid-sized companies often lack time and specialists to turn a tool into a productive solution. At the same time, contracts and customer data must not go into public models without control.
02
Vendors sell the platform
The vendor roadmap is not your bottleneck. Another licence does not change the process.
03
Data protection stalls experiments
Microsoft 365 is there, sensitive content too. Without a clear architecture, AI stays a demo nobody will approve.
Use
Where AI can be used in mid-sized companies
AI pays off especially where people regularly search for information, process documents or move data between systems.
Knowledge management
Find information from SharePoint, documents and internal stores faster. An AI knowledge base for companies answers and points to the underlying sources.
Process automation
Automate recurring flows such as invoice intake, document processing or data sync. AI is used where unstructured content must be processed.
Document processing
Classify emails, PDFs and other files, extract data and pass them to downstream processes.
Microsoft 365 and SharePoint
Use existing Microsoft 365 environments as the base for AI solutions. What matters is technical integration, permissions and controlled handling of company data.
Custom AI assistants
Build AI assistants for recurring questions and tasks, with connected sources and defined processes. Not a generic chat window without context.
The solution
One use case that pays off. Then the next.
A typical start is one of two places: finding knowledge in SharePoint and files, or automating a painful process with media breaks. I build both so existing systems stay and the solution is handed over with documentation.
Experience from insurance, logistics and regulated settings. Industrial engineering plus delivery: analysis and code, not slides someone else is meant to build. Custom AI delivery from €1,100 per day. The first conversation is non-binding.
What you actually get
- Clarify bottleneck, data risk and success criterion before building
- AI knowledge base or process automation as the first productive step
- Connection to Microsoft 365, APIs or mixed landscapes
- Model choice by fit: Azure, EU cloud or local models
- Documentation and handover to your team
Microsoft 365
When is Microsoft Copilot enough and when do you need a custom AI solution?
Copilot can be useful for certain tasks. If the real problem is a custom process, a complex data source or an integration between systems, a standard tool often is not enough.
Then you need a solution tied to your sources, permissions and workflows. Not another subscription next to Microsoft 365.
Which AI solution fits your company data?
- Public models only where no confidential company data is processed
- Microsoft Azure in the EU when cloud and contract fit
- Local models when data must not leave the company
- Access control, roles and logging from the start
- Data sources and permissions decide which information the AI may process
In practice
Three entry points that work in mid-sized companies
01
AI knowledge base for companies
People ask questions and get answers from SharePoint and files. With a source link and without putting data into a public store.
02
Automate one workflow
Invoice intake, mail or data sync: one clear trigger, less manual work. AI only for unstructured parts.
03
Orientation before budget
In the first conversation I clarify data, oversight and value before a project is set up.
04
Integration without a system swap
CRM, ERP and filing stay. What gets connected is the flow in between.
Questions
Common questions about AI in mid-sized companies
Is AI worthwhile without an in-house AI team?
Yes, if the first use case is concrete and the solution is handed over with documentation. I build so operations can be taken over in-house.
How does AI work with Microsoft 365?
Via Microsoft Graph, SharePoint, Outlook and existing permissions. Copilot is one option. A custom connection makes sense when the process or data source is more specific.
What does an AI rollout cost?
I work on a day rate. Custom AI delivery starts at €1,100 per day. Effort depends on the process, the system landscape and the data risk. The first conversation is non-binding.
How do I start without buying a platform?
With one concrete bottleneck: knowledge search or a flow with media breaks. Existing systems stay. What gets built is the connection in between.
Check where the leverage is, with no obligation
15 minutes is enough for a first reading: knowledge search or automation. After that it is clear whether and how to continue.