Microsoft 365
Automation and integration around Outlook, SharePoint, Teams, Excel and other parts of the Microsoft environment.
Technology
The right technical basis for your process
I do not work with a fixed platform. Depending on the task I combine existing tools, cloud services, local AI and custom software.
n8n can be the right orchestrator. Microsoft 365 can already provide the needed building blocks. A local model can keep sensitive data in the company. If standard setups are not enough, I develop the software that is needed.
What counts is not the tool, but the architecture behind it.
Which technical basis makes sense depends on the concrete process. Delivery ranges from classic automation and system integration through to AI applications.
Platforms
Use existing platforms where they fit
Many requirements can be solved with platforms and services already in place. If a company already uses Microsoft 365, Azure or other systems, I first check which of them can be used in a sensible way.
This is not about replacing an existing landscape with a new platform. Often more is gained by connecting what is already there cleanly and extending it with a clear purpose.
n8n
n8n connects APIs, databases, emails, AI models and existing systems in automated workflows. The platform is especially useful when several applications need to talk to each other and processes have to be orchestrated.
Cloud APIs
OpenAI, Anthropic, Google and other vendors can be integrated into existing applications and workflows via APIs. What counts is not only model performance, but also cost, data processing, region and the requirements of the concrete process.
AI
Use AI where it actually helps
Not every process needs a language model.
For structured rules, classic automation is often more reliable. AI becomes interesting when texts, documents, speech or other unstructured information have to be processed.
Depending on data risk and task, the model can run locally or via an enterprise API.
Local AI
If sensitive data should not leave your own infrastructure, a local model can be the better basis.
More on local AICloud AI
If stronger models, complex tasks or high load matter more, an enterprise API can be the fitting choice.
More on cloud AI
Software
When standard setups are not enough
Not every process can be pressed into an existing platform in a sensible way.
If requirements, interfaces or business logic demand it, I develop custom software and interfaces of my own.
The goal is not the largest possible technical landscape. Each component should do exactly the job it is suited for.
A setup can consist of several layers
- existing company systems
- APIs and databases
- n8n or another workflow engine
- custom software
- AI models
- local or cloud-based infrastructure
Architecture
Technology without unnecessary vendor dependence
Models and platforms change quickly. A good architecture should therefore not depend on one vendor remaining the best long term.
Where it makes sense, I separate application, workflow and model. A workflow can then keep running, for example, if the AI model in use is swapped later.
That reduces unnecessary vendor dependence and makes later changes easier.
Tools
What I typically use
Automation
n8n, APIs, webhooks and custom workflows
Microsoft
Microsoft 365, SharePoint, Azure and adjacent services
AI
Cloud models from different vendors as well as local open-weight models
Development
Custom software, scripts, interfaces and database connections
Infrastructure
Cloud, own servers and local systems, depending on data risk and requirements
The concrete choice comes only after analysing the process.
From platform to process
The question is not Which AI tool should I use?
Instead What should run automatically, reliably and traceably in the end?
Only then does the technical architecture follow.