Autopoiesis taps Oracle Cloud Infrastructure to build AI “co-scientist”
The startup is developing an AI model that combines state-of-the-art reasoning with systematic doubt to accelerate scientific discovery.
United States | Life Sciences
“When we were evaluating providers, Oracle’s team took time to understand our mission, our technical requirements, and our long-term vision. They genuinely cared about what we’re building.”
Standard AI models are like your mother: They tend to love everything you do. Which is great if you’re a kid, but frustrating if you’re any sort of scientific researcher.
Autopoiesis is on a mission to empower medical researchers with an advanced AI tool that doesn’t comfort their hypotheses but challenges them instead. The company picked Oracle Cloud Infrastructure (OCI) to deliver its unique “self-skeptical” AI model because of OCI’s raw performance, Oracle’s experience in healthcare and life sciences, and its close partnership with high performance chipmaker NVIDIA, among other reasons. Autopoiesis now provides medical and biological researchers with objective feedback on the likely outcomes of their theories, helping some of the world’s best minds save time, avoid mistakes, and make discoveries that could transform patient care and public health.
Why Autopoiesis chose Oracle
Autopoiesis isn’t shy about its mission: “To define the AI-for-science era and build the most impactful company in history. By enabling the world’s best scientists to move faster, we believe we can help save countless lives,” says Eike Gerhardt, the company’s cofounder and chief business officer.
Autopoiesis chose to run its “co-scientist” AI model, called Aristotle, on Oracle Cloud Infrastructure for several reasons:
- OCI’s GPU infrastructure and bare metal compute options provide the raw performance required for intensive model training. “Oracle has clearly designed its cloud infrastructure for performance-intensive workloads, and the results are evident,” CEO Joseph Reth says.
- OCI's RDMA cluster networking over a converged Ethernet interconnect (Oracle Acceleron) provides throughput for distributed training, which is critical when you're training large models across multiple nodes.
- Autopoiesis can scale its OCI capacity up or down easily for specific workloads for defined periods of time.
- Oracle’s very close working relationship with chipmaker NVIDIA gives Autopoiesis access to the latest hardware configurations, optimized for AI workloads. “When you’re pushing the boundaries of what's possible in AI-driven discovery, having that ecosystem advantage matters enormously,” Reth says.
Other factors that influenced Autopoiesis’s decision to choose OCI include Oracle’s deep experience in the healthcare and life sciences sectors, its experience working with companies in other regulated industries, and OCI’s rigorous security and regulatory compliance capabilities.
Results
Autopoiesis requires enormous computational resources to train its large-scale AI models for scientific discovery. It’s looking to OCI to reduce model training time and inference latency for scientific predictions, as well as for IT cost efficiency.
Autopoiesis cofounder Eike Gerhardt feels a deeply personal sense of mission in building a company at the intersection of advanced AI, science, and healthcare. He lost his father to cancer when he was only 12 years old, and his mother has twice battled cancer.
Already he’s seen the impact the company’s AI model can have. Reth, the company’s CEO, learned that the mother of a close friend was hospitalized with cancer. Although she suffered from noticeable breathing difficulties, doctors dismissed them as normal side effects, and even ChatGPT confirmed their assessment. But when Reth ran the data through Aristotle, Autopoiesis’s advanced AI model, it raised an urgent alert: acute respiratory distress, requiring immediate intervention.
The daughter rushed to the hospital and insisted that her mother receive emergency attention. It was later discovered that without that timely action, her mother would have suffocated within 24 hours. The episode underscored for the team that developing an AI co-scientist could save countless lives in the future by helping caregivers make better decisions when the stakes are highest.
About the company
Autopoiesis is a San Francisco–based startup building an AI model designed to help scientists accelerate discoveries, validate complex hypotheses, and address urgent challenges in medicine and biology.