After Mythos, Claude enters drug discovery race with ex-Google CEO startup help

SandboxAQ has partnered with Anthropic to bring advanced drug discovery and materials science tools to users through natural language. Here's how AI is now helping solve scientific challenges beyond automation.

In Short

• SandboxAQ brings drug discovery tools directly into Claude AI chatbot
• AI models can predict molecular behaviour before lab-based experiments begin
• Companies increasingly use AI to solve science and medicine challenges

As artificial intelligence continues to improve rapidly, many people fear that AI systems could replace human jobs across industries. But at the same time, advancements in AI are also helping researchers tackle scientific and medical challenges that humanity has struggled with for generations. Recently, Meta unveiled a new TRIBE v2 (Trimodal Brain Encoder) foundation model designed to predict how the human brain responds to almost any sight or sound. The model aims to help neuroscientists conduct experiments more efficiently and better understand brain activity. Now, another AI company, SandboxAQ has now partnered with Anthropic to integrate its scientific AI models directly into Claude, Anthropic’s AI assistant.

SandboxAQ is chaired by former Google CEO Eric Schmidt and has raised more than $950 million from investors. Its partnership with Anthropic allows researchers and scientists to access advanced drug discovery and materials science tools through a conversational interface using natural language. Through this partnership, researchers will no longer need specialised computing infrastructure or highly technical systems to run these models.

What are SandboxAQ scientific AI models ?

SandboxAQ has developed what it calls large quantitative models, or LQMs. Unlike traditional AI models that are largely trained to identify patterns in text, these models are “physics-grounded,” meaning they are built using the rules of the physical world.

The models can perform quantum chemistry calculations and simulate molecular dynamics and microkinetics — the process of understanding how chemical reactions unfold at the molecular level. In simple terms, the technology helps researchers predict how molecules are likely to behave before actual laboratory experiments begin.

This could help reduce the time, cost and complexity involved in discovering new drugs or materials.

The models are trained using real-world laboratory data and scientific equations, making them useful for scientists working at pharmaceutical and industrial companies searching for new materials or products.

Why is this significant?

According to Nadia Harhen via TechCrunch, SandboxAQ’s general manager of AI simulation, this marks the first time a frontier quantitative model has been integrated with a frontier large language model in a way that users can access naturally through conversation.

Previously, researchers using SandboxAQ’s LQMs needed to set up and manage their own digital infrastructure to operate the models.

- Ends