Prepare the work
AI summarises a case, organises open points and prepares the next step. A person reviews and decides.
AI for everyday work
I develop AI solutions that support people in concrete tasks: they bring together information from existing systems, prepare cases and expose gaps. A solid, vendor-independent architecture keeps sources, ownership and approvals clear.
My approach
Good AI does not take responsibility away from people. It takes away search work and prepares decisions.
It works with approved sources, exposes missing information and improves through corrections. A new rule applies only after the responsible person approves it.
A workday with AI
How AI supports everyday work.
A customer service administrator needs to know whether the appointment is confirmed, the spare part is ready and what the technician found last time. AI brings the facts together, exposes the gap and prepares the reply. A person still decides and sends it.
Read the workdayFrom tool to working system
Useful in one case. Reliable because of its architecture.
A chat can answer a question. At work, AI must also know which sources apply, what data it may access and when a person must decide.
Connect systems
It retrieves appointments, orders, feedback and instructions from the systems that already own them.
Learn with control
People correct the individual case. A reusable rule is created only after the responsible person approves it.
The foundation
Clear sources, ownership and approvals.
The architecture connects daily tasks with company processes and goals without mixing responsibilities.
Strategic knowledgeDirection, priorities, boundariesOwner: leadership
Process knowledgeHandoffs, rules, controlsOwner: process lead
Task knowledgeCases, instructions, exceptionsOwner: specialist team
The whitepaper
How AI can work reliably in a company.
The whitepaper explains the architecture behind everyday work: data access, knowledge domains, ownership, human approval and controlled learning.
Evidence from practice
I use these principles in my own work every day.
Selected work across reliable systems, public technology and AI product development.
More than 25 successful projects across finance, retail, industry and tourism
Public work on Azure Speech, Whisper and YellowCube
Personadeck: initiated by me and developed with three partners into a working AI product
My knowledge and work system with clear sources, ownership and approvals
The first step
Examine one recurring workflow in concrete terms.
We clarify what work is done today, where the necessary information lives and where AI could provide useful support. Only then do we assess technology and architecture.
Start a conversation