We are proud to announce our most recent investment in Tentris, the deep-tech startup building a next-generation graph database for enterprise AI, founded on more than eight years of research at Paderborn University. Tentris has secured €2 million in financing: a €925,000 pre-seed round led by Bloomhaus Ventures, with participation from us, Vanagon Ventures our friends from āltitude, 10x Value Partners, and business angel Pascal Wichmann, alongside €1 million in public innovation funding (EXIST Transfer of Research) from the Federal Ministry for Economic Affairs and Energy of Germany.
At the core of the platform is TentrisDB, a high-performance, disk-based RDF graph database built on a compressed Hypertrie index and worst-case optimal multi-way joins. The result: up to 150x faster query execution and up to 1,000x lower RAM consumption than conventional graph database approaches — making complex graph analyses economically viable that were previously too slow, costly, or computationally infeasible. The technology is already being evaluated by Fortune 500 companies that have hit the physical limits of legacy graph database architectures, and recently won first place at the global Knowledge Graph Conference in New York.
Why we invested: AI needs a new foundation
We currently ask LLMs to do two things at once: speak and think. But LLMs are probabilistic machines. Ask them to reason over complex, domain-specific enterprise data — they hallucinate.
→ The answer lies in separating language from logic. The LLM becomes largely a translator. The actual thinking — verifying fact chains, deriving new knowledge deterministically — happens in a structured ground truth layer built on knowledge graphs.
→ Tentris takes the reasoning task away from the language model and makes answers mathematically traceable and verifiable. It also provides the memory layer autonomous AI agents need to communicate safely.
“Today's LLMs fabricate facts, can't show their reasoning, and lack a structured memory. As AI moves from chatbots toward world models, correctness becomes critical. Built on mathematical breakthroughs from years of research, Tentris breaks the RAM-cost barrier that made graph-powered AI unaffordable at scale. By giving foundation models a deterministic 'brain' that retrieves and reasons over trillions of interconnected facts, with every result traceable and verifiable, Tentris lays the foundation for World Models and Enterprise AI that are transparent, auditable, and trustworthy.” - Susanne Fromm
The newly secured capital will be used to further develop TentrisDB, expand the engineering and commercial teams, and accelerate market adoption. Tentris has already secured its first enterprise customers and is working with a growing pipeline of mid-cap organisations evaluating the technology.
About Tentris
Tentris is a Bielefeld-based deep-tech company building the data infrastructure for the next generation of enterprise AI. Founded in 2024 as a spinout from Paderborn University by Alexander Bigerl, Tobias Rebert, Nikolaos Karalis, and Prof. Dr. Axel-Cyrille Ngonga Ngomo, Tentris develops high-performance graph database technology with the ambition to build a category-defining database company from Europe. More: tentris.io
About Vanagon
Vanagon is a Munich-based venture capital firm, purpose-built for deep tech teams from pre-seed. The mission: to invest in the most groundbreaking deep tech companies of the next decade that use the European advantage to scale globally. The alpha strategy: moving teams up the curve through a close partner network of allied founders, corporates and co-investors. More: vanagon.vc
Bests
Axel, Sandro & Susanne
