Sponsored by |
 |
|
|
|
| |
| |
| |
Good morning, AI enthusiast. |
Anthropic just gave Claude a body. A new open standard called the Model Hardware Standard lets AI agents plug directly into lab equipment - microscopes, robotic arms, liquid handlers - and run experiments on their own, cutting integration time from weeks to minutes. |
If an AI model can run a lab overnight without a human checking in, how much of the slow, manual grind of science is actually about to disappear? And who decides which mistakes an autonomous agent is allowed to make on its own? |
Today in AI Brief: |
Anthropic's new standard lets Claude run lab hardware Z.ai reveals its secret stealth model, Ox Alpha Nvidia's record quarter has Jensen calling an inflection point |
|
|
|
|
|
| |
|
AI Insights. Real Growth. Higher GMV, Better Profits |
|
The difference between growing stores and stagnant ones isn't more effort. It's better insights. StoreClaw analyzes your Shopify and Amazon data, surfaces your biggest growth opportunities, and helps you increase GMV while protecting profit. Start free with bonus tokens. No credit card required. |
Try it FREE today |
| |
| |
| |
Claude Learns to Run Lab Robots |
In Brief: Anthropic released the Model Hardware Standard, a shared specification that lets Claude and other AI agents operate lab and manufacturing hardware - microscopes, liquid handlers, robotic arms - that normally can't talk to each other, cutting setup time from weeks to hours or minutes. |
The Details: |
Genentech, Carnegie Mellon, and quantum computing startup QuEra are among the first research partners, alongside industry backers like AWS, Tecan, and QIAGEN. At Carnegie Mellon, dose-response experiments that used to take weeks to set up now run in 8 hours - 3x faster than before. QuEra's agent-written laser recovery script hit a 99.3% success rate, versus 58% with the old method, and runs in just 6 seconds. |
Take Away: |
Turning a lab into something an AI can operate solo removes one of the last bottlenecks between "AI reasons about science" and "AI does science." If the standard catches on the way MCP did for software tools, expect autonomous experiments to become routine rather than a novelty. |
|
|
|
|
|
| |
|
| |
| |
| |
China's Z.ai Reveals Its Stealth Model |
In Brief: For six days, an anonymous free model called "Ox Alpha" topped OpenRouter's usage charts, beating out DeepSeek without anyone knowing who built it. On August 26, Chinese AI lab Z.ai confirmed it was behind the model, unveiling it publicly as GLM-5.3-Flash. |
The Details: |
The model runs on 320 billion total parameters with just 18 billion active, using a mixture-of-experts design that handles text, image, and video input, according to its model card. Z.ai priced it at $0.15 per million input tokens and $0.50 per million output tokens - a fraction of what premium labs charge - with a launch promo cutting that in half through September 9. Z.ai's Hong Kong-listed shares jumped 12% on the news, and the stealth run reportedly doubled as a real-world stress test for inference running on Chinese-made AI chips. |
Take Away: |
Z.ai didn't just ship a good model - it proved an anonymous, free stealth launch can out-compete the field before anyone knows the lab's name. Combined with the pricing, it's another sign Chinese labs are racing to make frontier-adjacent AI too cheap to ignore. |
|
|
|
|
|
| |
|
|
The best voice models now listen, adapt, and resolve too. |
|
Most CX platforms don't own the voice. They orchestrate a workflow, then call a third party for speech and transcription. Every hop adds latency, and latency is what turns a frustrated customer into a churned one. |
ElevenAgents is the opposite. Built on the voice models the market already builds on, it runs voice, transcription, chat, and reasoning in one vertically integrated pipeline. Responses come back in under 400 milliseconds and sound human, not synthetic. When a caller gets frustrated, the agent detects it and shifts tone in real time: calm, reassuring, patient. |
You keep full control. Plug in any LLM, connect tools, webhooks, and MCP servers, and ground every answer in your knowledge base. Launch in minutes, A/B test with Experiments, enforce Guardrails, and version every change. |
More resolved conversations, less infrastructure stitching. Pricing is transparent and flat at $0.08 per minute. |
Start building free |
| |
| |
| |
Nvidia's Record Quarter Signals an AI Inflection Point |
In Brief: Nvidia posted second-quarter revenue of $96.2 billion, up 106% year-over-year, with CEO Jensen Huang declaring that "AI has reached its inflection point." |
The Details: |
Data center revenue hit $89 billion, up 117% year-over-year, as demand from AI labs and cloud providers keeps outpacing supply. Huang said AI's tokens are now "productive and profitable," pointing to accelerating demand across new labs, frontier models, and the open-source ecosystem. Nvidia guided to $108 billion in third-quarter revenue and returned $26 billion to shareholders through buybacks and dividends this quarter. |
Take Away: |
When a company just clocked roughly $60 billion in quarterly profit and still says demand outstrips supply, "AI hype" isn't really the right word for what's happening. Every model release and coding agent story this year keeps translating into hard dollars for the company that supplies all of it. |
|
|
|
|
|
| |
|
| |
| |
| |
Everything else in AI |
Google added a Gemini assistant to Google Chat, letting Workspace users search Gmail, Drive, and Calendar or draft messages with a Ctrl+G shortcut. |
SoftBank is reportedly negotiating a majority stake in 1X Technologies at a $6 billion valuation, up from $820 million in January, to gain control of its $20,000 home robot NEO. |
University of Queensland engineered cyborg cockroach "paraborgs" fitted with cameras and remote-triggered injectors that delivered simulated medical aid with 95% accuracy in disaster-response tests. |
Anthropic unified Claude's memory across chat and Cowork, letting users view, edit, or delete anything the AI has saved about them. |
|
|
|
|
|
| |
|
|
|
