SandboxAQ integrates drug discovery models into Anthropic’s Claude
The partnership allows users to access physics-grounded large quantitative models via natural language, challenging competitors who focus solely on model architecture.
SandboxAQ has partnered with Anthropic to integrate its proprietary scientific AI models directly into the Claude conversational interface. The collaboration aims to eliminate the need for specialised computing infrastructure, enabling users to access drug discovery and materials science tools through natural language. This move marks a strategic pivot for the Alphabet spinout, which positions accessibility and user interface as the primary bottlenecks in scientific research, rather than model capability.
The integrated models, known as Large Quantitative Models (LQMs), are described as "physics-grounded" and are trained on real-world lab data and scientific equations. Unlike standard large language models that rely on text patterns, these LQMs simulate molecular dynamics and microkinetics to predict how candidate molecules behave. This capability allows computational scientists, research scientists, and experimentalists to run quantum chemistry calculations and assess chemical reactions without leaving the Claude interface.
Nadia Harhen, SandboxAQ’s general manager of AI simulation, stated that this partnership represents the first time a frontier quantitative model is accessible on a frontier large language model through natural language. Previously, users of SandboxAQ’s LQMs were required to provide their own digital infrastructure to run the models. The new integration targets customers at large pharmaceutical or industrial companies who have previously found that the complexity of their problems did not yield positive results when translated into existing software.
SandboxAQ characterises its LQMs as being engineered for the "quantitative economy," a sector valued at over $50 trillion that spans biopharma, financial services, energy, and advanced materials. The company, which was founded approximately five years ago and has raised more than $950 million from investors, counts Eric Schmidt, former CEO of Google, as its chairman. The firm operates multiple business lines, including a cybersecurity division, alongside its scientific AI initiatives.
While competitors such as Chai Discovery and Isomorphic Labs have focused primarily on building better scientific models, SandboxAQ is betting that the bigger obstacle is access. The company argues that its approach of embedding powerful tools into a conversational interface will allow a broader range of researchers to utilise advanced simulation capabilities. The integration is intended to streamline the workflow for those searching for new materials that can become marketable products, reducing the reliance on highly technical computing expertise.
