SandboxAQ just pulled its scientific AI models out of the specialist sandbox and dropped them straight into Claude. According to TechCrunch AI, the Alphabet spinout teamed up with Anthropic to put its drug discovery and materials science tools behind a conversational interface, killing the need for users to spin up their own digital infrastructure. This is significant because the bottleneck in drug discovery isn’t always the science. It’s who can actually run the software.
What Got Launched
The integration brings SandboxAQ’s large quantitative models (LQMs) into Claude. These aren’t standard LLMs trained on internet text. They’re “physics-grounded” models built on the rules of the physical world, trained on real-world lab data and scientific equations.
What the models can do:
- Run quantum chemistry calculations
- Simulate molecular dynamics
- Model microkinetics (how chemical reactions unfold at the molecular level)
- Predict how candidate molecules behave before researchers touch a lab bench
The headline shift: a researcher can now query these tools in plain English through Claude instead of provisioning compute and wrestling with specialized software.
Why The Interface Matters
Most AI drug discovery startups have aimed their tools at computational scientists who already know what they’re doing. SandboxAQ is taking a different bet. As Nadia Harhen, SandboxAQ’s general manager of AI simulation, told TechCrunch AI: “For the first time, we have a frontier [quantitative] model on a frontier LLM that someone can access in natural language.”
That framing puts SandboxAQ on a different track than rivals like Chai Discovery and Isomorphic Labs, which have poured resources into building better underlying models. SandboxAQ’s pitch is that the model quality is already there. The friction is access.
Harhen described the typical customer profile: computational scientists, research scientists, and experimentalists at big pharma and industrial firms. “Our customers come to us because they’ve tried all the other software out there, and the complexity of their problem is such that it didn’t work or didn’t yield positive results for them when that translation went to take place in the real world,” she said.
The Company Behind It
SandboxAQ spun out of Alphabet roughly five years ago. Eric Schmidt, Google’s former CEO, sits as chairman. The company has raised more than $950 million and runs multiple business lines, including cybersecurity.
LQMs are positioned as the company’s wedge into what it calls the “quantitative economy,” a $50+ trillion bucket that covers biopharma, financial services, energy, and advanced materials. That’s a deliberate framing. SandboxAQ isn’t selling another chatbot or coding sidekick. It’s targeting the industrial economy AI is supposed to remake.
What Stands Out
The move flips a common assumption. Plenty of AI tools assume the user is technical enough to handle complex software. SandboxAQ is betting that pulling drug discovery work into a conversational layer expands the user base beyond computational chemists to scientists who know biology but don’t write quantum chemistry code.
It’s also a notable distribution play for Anthropic. Claude becomes the front door to specialized scientific tooling, not just a general assistant. If physics-grounded models can ride on top of frontier LLMs this way, expect more domain-specific scientific integrations to follow, in materials, energy, and beyond.
Full details are available at the original TechCrunch AI report.