Our mission is to make molecular engineering precise, efficient and scalable. By integrating physics, AI and experimental learning, we enable partners to design, prioritize and iteratively optimize functional molecular systems.
| Design Layer | Reality Layer | Decision Layer |
|---|---|---|
| Agentic AI explores designs, orchestrates analyses and synthesizes evidence. | Physics-based simulations and domain models evaluate molecular behavior, energetics and function. | Expert-defined policies integrate evidence and balance performance, risk and uncertainty. |
We envision a future where targeted biologics are predictably engineered for their intended function, with human experts and agentic AI working together as trusted research partners.
Instead of automating isolated steps, we connect data generation, analytics, simulations, and machine learning into a continuously learning system. Each experiment improves the decision base; each model update guides the next experiment. Over time, discovery becomes scalable, reproducible, and intelligence-driven.
Biological data is noisy, heterogeneous, and context-dependent. AI trained on it inherits both its power and its blind spots. We envision systems that flag missing metadata, express uncertainty explicitly, and pair data-driven models with physics-based simulations for orthogonal validation. When both perspectives agree, confidence grows; when they disagree, uncertainty becomes visible and actionable.
Breakthroughs happen when experimental and computational work operate in a shared loop. AI-enabled, closed-loop labs turn disconnected steps into adaptive discovery cycles, where results inform models and models guide the next experiment. Better science starts with better understanding.
Real progress in biotech comes not from predictions alone, but from understanding why decisions are made. Smart hybrid models combine machine learning, physics-based simulation, and expert knowledge to reason about biological systems in a scientifically meaningful way. Not blackbox AI, but models that think, test, and learn like scientists.
Building academic collaborations at the intersection of computation, biology, and molecular engineering.
Extended collaboration through 2027, focused on VHH developability and engineering.
Harald Kolmar Research Group · TU Darmstadt
Working on AI-accelerated molecular simulations for VHHs, conjugates, and their interactions.
AG Kronberger · University of TübingenFresh from BIO International in San Diego, we are excited to share where xyna.bio is heading next. We will be at BioTechX in Basel from 06th to 08th of October 2026 with our own booth, continuing the conversations around neuro-physical modeling, VHHs and better biological decision-making. San Diego reinforced a clear message for us: AI in biotech needs to deliver quality, not just speed. We look forward to bringing that perspective to Basel and connecting with teams working at the intersection of strong science, data, and next-generation discovery. See you at BioTechX.
Read moreWednesday, July 8, 2026What we took away from #BIO2026 in San Diego: strong science, real manufacturing needs, and a maturing view on AI. A few themes clearly stood out: ADCs, conjugates, and lipid-based drug delivery were among the hottest topics, while lab automation continues to gain momentum. Antibodies remain the dominant biologics language, but VHHs deserve a much stronger position in that conversation and take up more and more space. One of the most valuable takeaways for us was the quality of conversations around AI. We saw strong interest in our neuro-physical approach, especially from teams looking for more intelligent modelling, triage, and design support. At the same time, there is understandable caution. Many companies have already had mixed experiences with AI-based discovery: faster processes, but not always better candidates. The message was clear: AI needs to deliver quality, not just speed. For xyna.bio, this reinforces why trust, scientific grounding, and VHH-specific expertise matter. Within the broader GIP Exyr GmbH and Xyna ecosystem, this principle is already reflected in initiatives such as the Catalyst project by xyna.com on trusted agentic AI for access and permissions management, where transparency, explainability, auditability, and clear governance boundaries are central. We see the same need in biotech: the opportunity is not to use AI as a label, but to build reliable & trusted, domain-specific tools that support better biological decision-making. Thank you to everyone we met in San Diego. We are excited to keep the conversations going.
Read moreFriday, July 3, 2026We’re excited to announce an upcoming collaboration between xyna.bio and AG Kronberger at Universitätsmedizin Eberhard-Karls Universität Tübingen.This partnership aims to explore new frontiers in VHH engineering:A central goal of the collaboration is to build an AI system that can dramatically accelerate molecular simulations for VHHs. This would allow VHH and conjugates and their interactions to be modeled and analyzed much faster than with traditional computational methodsWe believe that integrating advanced AI with deep scientific expertise will open powerful new possibilities for drug discovery and precision therapeutics.Looking forward to what we will build together
Read moreWednesday, May 13, 2026Three times - always a charm!We are proud to announce that xyna.bio is a sponsor of the BFH European Meet-Up for a third time this year.It is impressive to see how early collaboration shapes success in later stages of the competition, and we look forward to engaging with the next generation of innovators.
Read moreTuesday, April 14, 2026
Use of xyna.bio is intended for research purposes only and not for
clinical decision-making. All in silico results, regardless of how
obtained, must be experimentally validated.
xyna.bio must not be used for the design, development, or facilitation
of biological weapons, harmful biological agents, or any illegal,
terrorist, or malicious activities.
