We integrate language models, RAG and AI agents into existing enterprise software, especially where privacy, regulation and years of grown code rule out the easy answers.
We advise, design and build it ourselves. The focus is on data, permissions and traceability, not on demo effects.
Language models inside your existing applications: search and answers grounded in your own documents, document analysis, and assistants with tightly limited data access.
Agent workflows that do real work: consolidating data, preparing decisions, triggering systems. With approval steps wherever a human has to decide.
We introduce agentic coding and spec-driven development to your engineering team: tools, workflows, quality assurance and clear boundaries.
Where does AI pay off in your system, and what stands in its way? We assess data, permissions, architecture and operations and deliver a prioritised roadmap.
The foundation for AI: step-by-step replacement of legacy systems without a big bang, cloud architecture on Azure and AWS, SaaS operations under regulatory requirements.
Interim technical leadership for teams without their own: ownership of architecture and roadmap, developer coaching and support with hiring.
Permissions, data residency and traceability are part of the architecture. European models and hosting in your own tenant on request.
We build for operations: monitoring, tests, cost control and a plan for the day the model gets it wrong.
From the architecture decision down to the code. Whoever advises also builds, and owns the result.
We hand over so your team can carry on without us: documented, tested and understood.
A selection of projects for large enterprises and the financial sector. We name clients only with their approval.
Contracts are analysed by AI and imported as structured data into the business application, for a governance, risk and compliance platform used by banks.
An assistant answers questions solely from the user manual and has no access to any other data, which keeps it usable in a regulated environment.
A company-wide knowledge base with searchable meeting transcripts. What was said in meetings and written in documents can be queried in one place.
A platform rebuilt entirely with agentic coding, using spec-driven workflows and clear quality gates for the AI agents.
Incremental migration of an entire frontend from Vaadin (Java) to React and TypeScript. Features move one at a time while regular releases continue.
A domain-driven microservice architecture for automated fuel pricing, followed by the migration of the whole platform from Azure to AWS within three months.
Every step delivers a result that stands on its own. After each one, you decide whether and how to continue.
Founder · Software architect
For more than twelve years, Sebastian has built and owned platforms for large enterprises and the financial sector. He has taken AI features in regulated products to production, led engineering teams of eleven, and run around 50 technical interviews.
prodigy.solutions brings that experience to companies that are serious about AI. Alongside client work, it builds its own products, including a framework for spec-driven, autonomous delivery pipelines and a multi-agent platform.
Don't see yours? Just ask us.
Wherever possible, in your own environment, for example your Azure tenant. We choose models and hosting to fit your requirements, including European providers such as Mistral on request. Permissions from your existing systems are respected by the AI as well.
In production we mostly work with Azure OpenAI, Mistral, Azure AI Search and Microsoft Foundry, and n8n for automation. We are not tied to any vendor and recommend what fits your use case, budget and regulatory context.
Yes, that is our focus. We integrate AI into grown systems and modernise step by step, without a big bang and without stopping ongoing operations.
Mostly remotely, across the German-speaking region and internationally. We are happy to come to you for workshops, especially in North Rhine-Westphalia. We work in German and English.
With a no-obligation intro call. Afterwards we propose a concrete first step, usually a short assessment or a fixed-scope prototype.
A no-obligation 30-minute intro call. You describe what you have in mind, and we tell you honestly what we think is feasible and worthwhile.