SandboxAQ brings its drug discovery models to Claude — no PhD in computing required

SandboxAQ Embeds Scientific AI into Anthropic's Claude

SandboxAQ, the Alphabet spinout chaired by former Google CEO Eric Schmidt, has partnered with Anthropic to integrate its specialized scientific AI models directly into the Claude chatbot. The move aims to address a critical bottleneck in AI-driven drug discovery and materials science: user accessibility. While many startups have focused on building more powerful models, SandboxAQ contends the primary hurdle for researchers is often the complex interface, and is now offering its powerful simulation tools through a conversational front-end that requires no specialized computing infrastructure from the user.

Physics-Grounded Models Meet Conversational AI

With over $950 million in funding, SandboxAQ has developed proprietary large quantitative models (LQMs) that differ fundamentally from traditional large language models. These models are engineered based on the rules of the physical world rather than patterns in text, allowing them to perform high-fidelity simulations for researchers in biopharma, energy, and advanced materials. Previously, accessing these tools required significant computational expertise and infrastructure.

• Model Type: Large quantitative models (LQMs)
• Foundation: Built on scientific equations and real-world lab data, making them “physics-grounded.”
• Core Capabilities: Can run quantum chemistry calculations, simulate molecular dynamics, and model microkinetics to predict how chemical reactions will occur.
• Target User: Computational scientists, research scientists, and experimentalists at large pharmaceutical and industrial companies.

Expanding Access Beyond Computational Experts

This integration marks a strategic pivot from improving model capability to improving model usability. By embedding LQMs within Claude, SandboxAQ aims to empower a broader set of scientists who may not be computational experts but need to understand how candidate molecules will behave. “For the first time, we have a frontier [quantitative] model on a frontier LLM that someone can access in natural language,” said Nadia Harhen, SandboxAQ’s general manager of AI simulation. This approach differentiates the company from other well-funded ventures like Isomorphic Labs, which have concentrated primarily on advancing the underlying science of the models themselves.

By embedding its complex quantitative models into a conversational LLM, SandboxAQ is betting that the next major value unlock in enterprise AI lies not in model performance alone, but in abstracting away computational complexity to broaden the user base.