Keep pulling the thread on Sam Rodriguez.
In May 2025, Edison Scientific's multi-agent system, Robin, generated a new hypothesis for treating dry age-related macular degeneration, which was subsequently validated in wet lab and animal experiments.
A paper detailing a discovery made by Edison Scientific's AI agent system, Robin, was published in the journal Nature.
The original version of the Cosmos AI agent could run for 6 to 12 hours, write 45,000 lines of code, and read 1,500 papers in a single execution.
DeepMind's AlphaFold is a notable exception to the history of AI overpromising in drug discovery, as it has significantly changed molecular design by providing protein structures.
Edison Scientific's strategy includes training models on proprietary internal datasets from pharmaceutical companies, providing them with a sustainable competitive advantage that large, general model providers like OpenAI and Anthropic do not currently offer.
Edison Scientific's first multi-agent system, named Robin, generated a new hypothesis for treating age-related macular degeneration, which was subsequently published in the journal Nature.
Since its launch, the Cosmos AI agent has been used to make approximately 20,000 to 30,000 novel scientific findings.
The complete wiring diagram of the human brain has not yet been mapped.
Sam Rodriguez estimates that users of the Cosmos AI agent system have made approximately 20,000 to 30,000 novel scientific findings.
Sam Rodriguez predicts that future pharmaceutical companies will operate with much leaner teams due to AI, enabling them to pursue significantly more drug programs in parallel with the same number of employees.
Dry age-related macular degeneration affects approximately 5% to 10% of people over the age of 50.
The Cosmos AI agent system includes a built-in orchestrator for self-orchestration and uses 'world models' for sophisticated context management over hundreds of subagent runs.