The second startup
Emil Woermann, Jacob Leffers, Benjamin Seifert and Santiago Martinez-Avial were not first-time founders when they started Logicc in Hamburg in 2025. The four had previously built OAK25, a startup they eventually sold to a competitor. That exit taught them something most founders learn too late: the unsexy problems are the valuable ones.
Their second act began with an observation that was hiding in plain sight. By 2025, every professional in Germany knew about ChatGPT, Claude and Gemini. Lawyers, doctors, tax advisors and public institutions wanted to use them. But they couldn't — not legally. Germany's strict data protection regime, including GDPR and Section 203 of the German Criminal Code covering professional secrecy, made pasting client data into a US-hosted chatbot a compliance violation waiting to happen.
The big AI labs weren't going to solve this. Their business model depends on training on user data and hosting on their own infrastructure. The four founders realized that the compliance barrier wasn't a bug in the market — it was the market. Someone needed to build the AI interface that regulated professionals could actually use without risking their license.
Compliance as the product
Logicc's founding insight was that security couldn't be a feature bolted onto a generic AI tool. It had to be the foundation. The team built the platform from the ground up with what they describe as a zero-knowledge architecture: hosted exclusively on German servers, with zero data retention and full GDPR compliance baked into every layer.
This wasn't just marketing language. For a lawyer bound by professional secrecy or a doctor handling patient data, the difference between 'we take privacy seriously' and 'we architecturally cannot see your data' is the difference between usable and unusable. Logicc made that distinction its entire positioning.
The product itself is deliberately familiar: a single secure interface that aggregates AI models — ChatGPT, Claude, Gemini and others — with industry-specific functions, custom AI assistants and team collaboration features. The innovation isn't in the models. It's in the wrapper that makes them legally safe for the professions that need them most. Over 1,800 organizations now rely on the platform, spanning law firms, medical practices, tax advisory firms and public institutions.
The approach also solved a practical problem that generic AI tools ignore: professional workflows. A lawyer doesn't just need a chatbot; they need matter-specific assistants, document-aware answers and team workspaces where junior associates and partners collaborate under the same compliance umbrella. Logicc's industry-specific functions and bespoke AI assistants were built for these realities, not retrofitted onto them. That depth is why organizations adopt it as infrastructure rather than experimenting with it as a toy.
The €2.5M bet
In April 2026, Logicc announced a €2.5 million seed round, a signal that investors saw what the founders saw: Europe's regulatory environment, often framed as a handicap for tech, was actually a moat. As the EU AI Act tightened requirements and GDPR enforcement continued, the demand for compliant AI infrastructure only grew.
Co-founder and CEO Emil Woermann was direct about the ambition. With the new capital, he said, the company had a clear goal: to tenfold revenue over the next twelve months. That kind of public target from a seed-stage CEO is either bravado or conviction. Given that the company reportedly reached €5M in ARR within roughly eighteen months of founding, the conviction seems to have been warranted.
The timing helped. 2026 saw continued momentum across Germany's secure AI market, with multiple enterprise-oriented AI startups raising rounds. But Logicc's wedge was sharper than most: rather than selling 'AI for everyone,' they sold 'AI for the people who are legally barred from using it.' That specificity made every sales conversation easier, because the prospect's pain was not hypothetical — it was regulatory.
Why it worked
Logicc's story is a case study in finding a market inside a constraint. Most AI startups in 2025 were racing to build the best general-purpose tool. The Logicc founders went the opposite direction: they picked the customers with the most restrictions and built specifically for them. The result was a product with a smaller total addressable market on paper, but dramatically higher willingness to pay and near-zero competition in its niche.
The second-time-founder advantage mattered too. Having sold OAK25, the team knew how to ship, how to sell to businesses, and — critically — how to avoid the mistakes that kill first-time enterprise startups. They didn't spend a year finding product-market fit; they started with a compliance gap they could describe in one sentence and built directly toward it.
There is a broader lesson here for founders watching the AI wave. The model layer is commoditizing fast. The durable value is in the last mile: the industry-specific, regulation-aware, workflow-integrated layer where generic tools can't go. Logicc didn't build a better model. They built the only door through which an entire class of professionals could walk, and they charge admission.
For founders, the takeaway is uncomfortable but useful: the biggest opportunities in AI may not be in building smarter models, but in building the trust layer that lets the most cautious buyers use them. Regulation is slow, boring and unglamorous — which is exactly why it keeps competitors out and customers in.
The playbook
- Sell to the constrained — Instead of competing for the broadest market, Logicc targeted professionals who were legally barred from using generic AI — turning regulation from a headwind into a moat.
- Make compliance architectural, not cosmetic — Zero-knowledge architecture and German-only hosting weren't marketing claims; they were the product. For regulated buyers, 'we can't see your data' beats 'we take privacy seriously.'
- Aggregate, don't invent — Logicc didn't train models. It wrapped ChatGPT, Claude and Gemini in a compliant interface — capturing value at the distribution layer while the labs fought over the model layer.
- Use founder experience as a weapon — Having sold OAK25, the team skipped the discovery phase most startups burn a year on. They started with a one-sentence problem and built straight at it.
- Set public revenue targets — Announcing a 10x revenue goal alongside the seed round created accountability and signaled confidence — a forcing function that aligned the whole company around growth.