Fetching from the wire…
Public story · 2026-07-02 · high
Nobel laureate John Jumper reportedly left DeepMind for the effort, per MIT Technology Review.
Why now: MIT Technology Review's June 30 report is what put the launch and Jumper's reported exit from DeepMind in the same story.
Anthropic launched Claude Science on June 30, aiming to automate computational biology and drug development the way Claude Code automates programming, per MIT Technology Review.
For anyone betting on the agents-for-a-domain thesis, this is the harder test: biology has real experiments to survive, not a compiler.
Drug discovery has more structured, simulate-able steps than most engineering work, which makes it a plausible next domain. It's also a harder one to fake: computational-biology results get checked against reality in ways generated code sometimes isn't.
Coverage ties the launch to a bigger name: Nobel laureate John Jumper, the scientist behind AlphaFold, reportedly leaving Google DeepMind for Anthropic. The report doesn't say whether either company has confirmed the move. What it does show is timing: the reported exit and the Claude Science launch landed close enough that MIT Technology Review covered them together.
Nothing in the launch names a drug candidate or a published result yet. That's the gap between announcing an agent for a domain and proving one works. Claude Code had to close that gap before anyone trusted it with real code.
Each link below shares sources, entities, or timing with this story.
The beta workbench (macOS and Linux, Pro/Max/Team/Enterprise) acts as a project manager across 60-plus scientific databases, rendering 3D protein structures, genome browser tracks, and chemistry drawings alongside reproducible code. Anthropic explicitly said it is "not a new A...
MIT Technology Review covers a Stanford/MIT effort (Anka Reuel, Shayne Longpre) analyzing 24,521 donated conversations across 52 models from 2023-2025 against vendors' own usage reports. Because Anthropic filters for work-related use, nearly half of real conversations would be...
Three frontier models shipped in a single week this month, and teams with a standing eval harness had a routing decision in hours. Anthropic's own agent-eval guidance says 20-50 tasks drawn from your real usage and real failures is enough to detect issues (DeepEval). DeepEval...
Moonshot AI released Kimi K3, a sparse mixture-of-experts activating 16 of 896 experts per token. That's about 1.8% of the pool live at any moment, with a 1M-token context window and native vision. Two new architectural pieces show up: Kimi Delta Attention and Attention Residu...
The IDE market is fragmenting, and this week drew the sharpest lines yet. Cursor 3 launched as a rebuilt agent-orchestration platform in Rust and TypeScript, replacing the VS Code fork with an Agents Window for dispatching and monitoring multiple AI coding agents. Anysphere hi...
This one hit different because I use Claude Code every single day. Anthropic published a disclosure confirming that a Chinese state-sponsored group weaponized Claude Code to conduct autonomous cyber espionage against roughly 30 targets, including tech companies, financial inst...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.