aiminute. ← All AI news
Tools AI Minute Newsroom 2026-08-28

The protocol that let AI touch your files now has a sibling for lab equipment — Anthropic wants a model to drive a microscope the way it drives an app

The protocol that let AI touch your files now has a sibling for lab equipment — Anthropic wants a model to drive a microscope the way it drives an app

Anthropic published a research preview of the Model Hardware Standard on Wednesday 27 August 2026, a specification meant to do for physical instruments what the Model Context Protocol did for software tools. Today, setting up and integrating equipment in a research lab or a factory takes weeks or months, because most devices cannot talk to each other and a specialist has to hand-build a translator for every pairing. MHS defines a common driver layer that exposes any device with a programmable interface through simple primitives — read, write — makes it discoverable in a uniform format, and attaches natural-language tags describing what the hardware is and what it can do, so a model can work out how to use an instrument without a bespoke integration written for it. It is model-agnostic and designed to sit alongside standard protocols including MCP, and Anthropic says integration work that took weeks can drop to hours or minutes. The work began with the HHMI Janelia Research Campus; the preview partners include Genentech, the University of Washington, Carnegie Mellon, QuEra Computing and Tetsuwan Scientific, alongside hardware vendors such as Amazon Web Services, Tecan, QIAGEN and Doosan Robotics. Access is limited for now to scientific labs and advanced manufacturers, who can apply at modelhardwarestandard.com.

Why it mattersFor two years the interesting question about AI in science has been whether a model that can read every paper can also run an experiment. The bottleneck was never the reasoning — it was that a pipetting robot, a microscope and a plate reader each speak a private dialect, and gluing them together is a job for a human engineer. A shared driver layer is the unglamorous piece that makes the rest possible, and MCP is the precedent: it was also a dull specification, and within a year it was how almost every agent reached the outside world.
#AI Agents#Robotics#Science & Research

✓ Verified · 2 sources

WhatsApp X Telegram
Read in the app — free, in 9 languages

Related stories

Google Search will now watch flight prices for you and book the hotel — the conversation ends at a booking, not a list of links
2026-08-28
The labs whose models broke out and hacked people have now co-signed a letter asking the world to hurry up and defend itself
2026-08-28
A 25-centimetre robot that walks, kicks and roller-skates, for $399 — and every behaviour on it can be retrained
2026-08-27
Google has put a hop in front of every link in its results. The stated point is to make its results unreadable to machines.
2026-08-27
A German broker with €60 billion now lets ChatGPT and Claude place your trades. You still have to press yes.
2026-08-27