kausable
A Heidelberg AI lab founded in 2025 researching alternatives to repeated model retraining, developing systems grounded in cause-and-effect relationships and synthetic causal data.
Europe's most interesting startup stories. Every week.
No spam. Unsubscribe anytime.
Podcast Episode
🎙️ Deep Dive
telli, kausable, Yope — $15M for AI agents, €12M for adaptive models and $12.3M for private social — Daily News July 23, 2026
Listen to the episodeAbout kausable
kausable was founded in 2025 in Heidelberg by Johannes Haux (CEO), Dr. Benjamin Herdeanu and Gregor Ramien, with the goal of rethinking how AI models learn and adapt. Continuous retraining makes many AI systems expensive and inflexible when environments change: kausable develops models grounded in cause-and-effect relationships and synthetic causal data to build systems that understand how environments change and apply knowledge to unfamiliar situations from a limited number of examples. With a team of 9, the startup closed a €12M seed round in July 2026, led by UVC Partners and Entourage, with participation from HTGF (High-Tech Gründerfonds) and Mätch VC. The funds will be used to expand the team, develop the models further and validate them with initial industrial customers in sectors including robotics, industrial systems, forecasting and healthcare. Synthetic causal data could reduce dependence on large historical datasets — one of the fundamental bottlenecks in developing adaptive AI.
The Story
Founded in 2025 in Heidelberg by Johannes Haux, Dr. Benjamin Herdeanu and Gregor Ramien to develop causal and adaptive AI.
How kausable works
Business Model
Development and future B2B commercialisation of foundation models and vertical applications; licensing and APIs.
Revenue Model
Licensing, APIs or industrial collaborations; the final commercial model has not been disclosed.
Products & Services
Key Products
- Reasoning-first frontier AI models
- Causal world models
Core Use Cases
- Robotics
- Industrial systems
- Forecasting
- Healthcare
- Model adaptation to new contexts
Market & Clients
Key Customers
Early pilot companies in industrial systems, robotics, forecasting and healthcare; names have not been disclosed.
Geographic Presence
How kausable competes
Competitive Advantages
- Cause-and-effect modelling approach — alternative to continuous fine-tuning
- Adaptation to new contexts from limited examples
- Synthetic causal data: reduces dependence on large historical datasets
How kausable grows
Growth Strategy
Expand the team, develop the models and validate them with initial industrial customers.
Distribution Model
Research partnerships, pilot projects and future distribution through licensing or APIs.
Moat (Defensibility)
Intellectual property on causal research, proprietary synthetic data and the team's scientific expertise.
Regulatory Context
European applications will be subject to the AI Act and to sector-specific rules for the use cases involved (healthcare, industrial).
Key Risks
- Technology remains research-stage — production transfer not yet demonstrated
- Generalisation challenges outside laboratory settings
- High compute requirements
- Competition from major AI laboratories with superior resources
Strategic Insights
- Synthetic causal data could reduce dependence on large historical datasets.
- The decisive test will be demonstrating adaptation in real industrial applications outside the laboratory.
Funding & Investors
Total Funding
Funding Rounds
- - Seed: 12000000
Lead: UVC Partners
Key Investors
Founding Team
Founders
Key Metrics
Lessons from kausable
- A strong scientific thesis should be connected early to measurable use cases.
- Fundamental research requires patient capital and technically capable investors: HTGF and UVC Partners are coherent profiles.
Similar Startups
Tower
Berlin startup founded in 2024 for Python data pipelines. In August 2026 MotherDuck announced it had acquired it.
Read AnalysisWeSort.AI
Würzburg company founded in 2022 that uses AI and X-rays to analyse and separate waste. In March 2026 it announced €10 million in funding.
Read AnalysisPackmatic
Berlin platform founded in 2021 for buying packaging. On 24 January 2024 it announced a €15 million Series A led by EQT Ventures.
Read Analysis