Summary
Max explains how CuspAI combines physics-informed neural networks, LLM orchestrators, and precise computational tools to search huge materials spaces and guide lab workflows. He cites a Finnish PFAS-removal project spanning 300,000,000,000,000 possible metal organic frameworks, then details the AI Materials Foundry’s role in shared datasets, partner labs, and cautious deployment with safety controls.
Key Topics
5 key topics from Max Welling and Larissa Holzki at HumanX Amsterdam 2026. Thicker branches were mentioned more often in the talk.
The Future of Materials Discovery
Click a topic to open it
Key Takeaways
- Use physics symmetries to make molecular AI models more accurate.
- Orchestrate literature search, generation, computation, and lab work through AI agents.
- Build shared datasets around pre-competitive materials problems to accelerate the field.
- Plan for device-level bottlenecks after synthesis and property validation.
- Control platform access and monitor requests to reduce dual-use risk.
“our platform has already shown it can much more efficiently guide the experiments towards the best material.”
Share
Up Next
FIZ
WatchNext in agenda
When the Barrier to Building Software Falls
Fabian Hedin, Iain Martin, Zivile Einikyte
agentic codingai startups
S
WatchAlso on ai agents
From Models to Agents at Scale
SSwami Sivasubramanian
ai agentsproduction ai
CN
WatchAlso on ai agents
Is AI Killing Software as a Service?
Clay Bavor, Niall Firth
ai agentssaas