What I've built, and what I owned.
Each role is written as a short case study: the context, what I was responsible for, and what came out of it. Where numbers are public, they're here. Where they aren't, I'm happy to walk through them in conversation.
-
AI Product Manager
Leading the agentic-AI initiative on Connectware, an industrial data platform that connects factory equipment to IT systems for major German industrial players.
- Shipped Connectware GPT 0→1: the platform's LLM assistant, now in public beta.
- Built the Cybus MCP server 0→1: an API surface that lets AI agents work with millions of live industrial data points.
- Own AI trust end to end: guardrails, governance, and output quality for every AI feature.
Context
Connectware runs in real factories. It moves millions of live data points between machines and business systems, and the customers who rely on it are large German industrial companies. That sets the bar for any AI feature: an answer that is probably right is not shippable.
What I owned
The agentic-AI initiative, from the first conversation about whether it made sense to the public beta. That includes problem framing and discovery with customers, the product definition, working day to day with engineering, and the governance work that decides what the AI is allowed to do and how we know its output is good.
Connectware GPT
An LLM assistant inside the platform that helps users understand, configure and troubleshoot their industrial data flows.
- Took it from concept to a working 0→1 release and into public beta.
- Defined the evaluation approach for output quality, so that “does it work?” has a measurable answer rather than a demo.
- Designed the guardrails: what the assistant may access, what it may change, and how it explains itself.
Cybus MCP server
A Model Context Protocol server that exposes Connectware’s capabilities and live data to AI agents in a controlled way.
- Built 0→1 as an API surface for agents rather than a chat feature, so that customers’ own agents and tools can work with factory data.
- Owned the permission and governance model: which data an agent can see, which actions it can take, and how that is audited.
What I learned
Trust is the product. Shipping the model was the easy part; deciding what it must never do, and proving it, is where the real product work was.
-
Product Manager, HealthTech
Product manager at wellabe, a Munich HealthTech company, working on digital health products where user trust and regulatory care shape every decision.
Context
wellabe builds digital health products. Working in HealthTech meant that user trust, data protection and clinical care were product requirements, not afterthoughts. That is also where my habit of treating “trustworthiness” as a first-class feature started.
What I owned
Product management for the digital product, from discovery with users and partners through delivery with the engineering team.
Full write-up in progress.
M.Sc. Management
Business and technology, with a focus on product and innovation.
Major in Artificial Intelligence
The quantitative foundation behind how I evaluate models and frame product problems.