OceanSparx develops AegisMind, a multi-model AI research engine, and publishes the discoveries that survive adversarial debate and formal verification at solver.press.
An AI research company focused on scientific discovery
OceanSparx's active work is AI-driven research discovery. We build AegisMind, which generates, debates and formally verifies cross-domain scientific hypotheses, and we publish the results openly at solver.press.
AI can produce plausible-sounding science faster than anyone can check it. We treat verification as the product: every claim is challenged by independent models and tested with formal methods, and what does not survive is not published.
Each published discovery comes with an Experimental Validation Package covering methodology, cost estimates and dependency maps, so research teams can see what it would take to test the claim themselves.
A multi-model AI research and discovery platform
Five frontier models (Claude, GPT, Gemini, Grok and Mistral) each generate hypotheses independently.
The models critique one another's hypotheses in structured debate, exposing weak assumptions.
Surviving claims are checked with the Z3 theorem prover before they are considered for publication.
Discoveries that pass every stage are published at solver.press.
Publication & discovery site
solver.press is the public window into the AegisMind discovery engine. It publishes cross-domain scientific hypotheses generated, debated and formally validated by the AI ensemble.
Research teams can access full Experimental Validation Packages, including methodology, cost estimates and dependency maps, for any published discovery.
Interested in our research, or in collaborating on AegisMind and solver.press discoveries? Send us a message.