“We have a strong use case.”
But nobody can say with confidence whether RAG, an agent, a workflow, conventional software or an existing product is the right answer.
AI Solution Architect & Product Builder · Austria / DACH
A good AI idea eventually has to become a working system.
I take AI initiatives from idea to production, connecting business, product, architecture and engineering.
Without a large project team. Without unnecessary technology baggage.
OpenAI Partner Network · Select Partner
The gap between idea and operations
AI Architecture and Product Development for Businesses.
From use-case assessment through AI architecture, RAG, agents and integration to a production-ready AI system. The decisive factor is whether business, data, UX, security and operations work together.
“The demo works.”
But not yet with real permissions, real data, stable evaluation and clean integration.
“We want to stay independent.”
Cloud, EU hosting, open source and on-premise must make technical and economic sense together.
This is where I work.
I help companies decide what AI can do and turn that decision into a working product or system.
How I work
From AI architecture to production: strategy is only the beginning.
01 · UNDERSTAND
Solve the right problem.
Clarify the business goal, users, data and existing systems before choosing technology.
- Use case
- Processes
- Users
- Data
- Business case
02 · DECIDE
Choose an AI architecture that holds up in production.
Define models, data flows, integrations, permissions, hosting and technical boundaries.
- AI architecture
- Make or buy
- Security
- APIs
- Evaluation
03 · BUILD
Test an AI pilot with real software and real data.
A focused pilot with real users instead of months spent on concepts.
- Prototype
- MVP
- RAG
- Agents
- Automation
- Integration
04 · PUT INTO PRODUCTION
Turn the demo into a production-ready AI system.
Measure quality, handle failures, establish operations and hand over cleanly.
- Production readiness
- Monitoring
- Guardrails
- Operations
One point of contact across the whole journey.
Why Thomas
One problem. Four perspectives. One person accountable.
AI projects often split early across strategy, UX, software development, AI engineering and infrastructure. Context, speed and accountability get lost in the handoffs.
BUSINESS
Value first.
Start with the problem and commercial value, not the model.
PRODUCT
People have to use it.
Impressive AI without understandable UX creates no value.
ARCHITECTURE
The whole system has to hold.
Data, APIs, models, security, hosting and operations are considered together.
ENGINEERING
I can build it myself.
Architecture decisions are implemented and tested, not just documented.
Fewer handoffs. Faster decisions. Systems that work beyond the architecture diagram.
Selected work
Experience made visible in working systems.
experdoo GmbH
Private AI platform in a regulated environment
- Situation
- Sensitive company knowledge and regulatory requirements ruled out uncontrolled public-cloud AI.
- Decision
- A self-hosted architecture with clear separation of data and permissions.
- System
- RAG for internal knowledge and agents for research and summarisation.
- Outcome
- Production AI without public-cloud dependence; positive internal security and architecture reviews.
- Role
- Architecture and implementation as AI Solution Architect.
Client project
Semantic matching for product catalogues
- Situation
- Web content needed to be matched with relevant offers in a large product catalogue.
- Decision
- Semantic matching of web content and products.
- System
- Web content was collected automatically and compared with products based on semantic similarity.
- Outcome
- The assessed matches were made available through an API.
- Role
- Co-development of the content collection and matching solution.
“Thomas grasped the complex project quickly, communicated clearly, and delivered reliably. That blend of technical understanding and professional collaboration makes the difference.”
Not from the AI hype
From 25+ years of digital practice.
I have developed digital products, platforms and systems for decades – for SMEs and international organisations, including in the Microsoft/MSN environment. More than 300 projects taught me that technology, users, UX, business model, data, architecture, integration and operations must work together. That experience helps me recognise patterns and risks early.
What interests me about AI today is not the next demo. It is the question: how does this become something that actually works for a company?More about Thomas
Architecture without dogma
Sovereignty where it brings a real advantage.
Privacy, EU hosting, open source and on-premise belong in the toolkit. The architecture follows actual requirements for data, security, cost, maintainability and independence.
Explore sovereign AIWays to work together
Four ways from a question to a working system.
The right starting point depends on where your project stands.
01
Explore all servicesAI Architecture & Build Sprint
For concrete AI ideas whose technical route is still unclear.
Outcome / focus
Architecture decision, make/buy, MVP scope and a concrete build plan.
View related service →02AI Pilot & Product Build
For teams testing a real workflow with real data and users.
Outcome / focus
A working pilot, a clear learning question and production-oriented implementation.
View related service →03AI Production Care
For AI systems already in production.
Outcome / focus
Evaluation, regression, RAG and agent quality, provider changes, cost and technical health checks.
Discuss this offer →04Fractional AI Technical Lead
For companies where AI matters strategically but a senior in-house AI team would be premature.
Outcome / focus
Technical leadership, architecture decisions and hands-on guidance.
Discuss this offer →Next step
Where is your AI project stuck?
Already have an AI idea, a pilot or a project that has stalled? Show me briefly what you are building. I will give you an honest technical and commercial assessment – including whether I am the right person to help.
Discuss an AI project