In partnership with
shared.image.missing_image
shared.image.missing_image
Good morning, {{first_name | AI enthusiast}}.
Agility Robotics just showed off Digit 5, a humanoid built to work the floor next to people without a safety cage around it — when someone gets too close, the robot simply stops, steps aside, or sits down.
With $300 million already in orders and its predecessor logging 65,000 hours at Amazon and Toyota warehouses, does ditching the cage finally make humanoid fleets practical at scale, or just move the risk closer to the people it’s meant to work beside?
Today in AI Brief:
Agility’s humanoid skips the safety cage
Claude now leads a quarter of its own R&D
OpenAI’s Astra tackles legal research
100 real-world tasks. 8 tool categories. 1 complete DevOps stack.
Master Git, Docker, Kubernetes, Linux, CI/CD, Terraform, and monitoring through structured tasks across real-job scenarios, and earn a shareable credential that proves you can apply what you’ve learned.
No lectures, no theory. Build a public portfolio as you go. Earn a verified badge. Free.
Agility’s Digit 5 Skips the Safety Cage
In Brief: Agility Robotics unveiled Digit 5, a humanoid built to work alongside people on warehouse and factory floors without the safety fencing today’s robots require.
The Details:
Digit 5 uses onboard sensors and AI to track nearby workers, and stops, steps aside, or sits down the moment someone gets too close.
The robot lifts 50 lbs repeatedly, reaches 7.2 feet, and now runs 20+ hours a day off a battery that recharges in just 9 minutes.
Agility says it already has $300 million in multi-year orders, with Digit’s predecessor logging 65,000+ hours at GXO, Schaeffler, Amazon, and Toyota sites.
Take Away:
Removing the cage is what actually lets a warehouse pack multiple humanoids next to human workers instead of walling each one off. Early access starts in the first half of 2027, with Agility’s Oregon factory built to eventually ship 10,000 units a year.
Anthropic Says Claude Now Leads 26% of Its Own R&D
In Brief: Anthropic published a new R&D Automation Index showing Claude now “leads” 26% of the company’s AI research and development work, up from under 1% in February.
The Details:
Claude organized roughly 15,000 sampled tasks into 542 categories, then rated how automated each one already is — the same model doing the measuring.
About 30,000 agents run at once on Anthropic’s main internal platform, and real-time monitors block just 1 in 47,000 of their decisions before they execute.
The company insists Claude isn’t operating fully autonomously on any task category yet, and only 6% of R&D compute currently goes to safety work.
Take Away:
Anthropic is trying to make the AI-building-AI feedback loop measurable instead of speculative, and a “leads” score of 26% within seven months of near-zero says the loop is already spinning. Whether the oversight numbers keep pace with that curve is the real story to watch.
SPONSORED BY TIGERDATA
Analytics on Live Data Without Leaving Postgres
When analytics on Postgres slows down, most teams add a second database. Then come the pipelines, the sync jobs, and a copy of your data that's always a little behind.
TimescaleDB takes a different approach: extend Postgres instead of splitting away from it. Hypertables partition your data automatically as volume grows. Hypercore compression cuts storage up to 95%. Continuous aggregates keep dashboards live without re-querying everything.
CERN runs Postgres this way for sensor data from the Large Hadron Collider.
No split architecture, no pipeline lag, no new query language to learn. Same SQL, same drivers, same tools.
Start on Tiger Cloud and get $1000 in credits.
OpenAI Rolls Out Astra for Law
In Brief: OpenAI introduced Astra for Law, a version of its GPT-6 Astra model paired with a legal search index spanning more than 230 million sources for law firms and legal tech vendors.
The Details:
The legal index covers U.S. case law, statutes, regulations, and court rules, with new sources added daily and 26 ecosystem plugins connecting ChatGPT to law-firm-specific tools.
Compared to GPT-6 Astra using plain web search, Astra for Law found 24% more reference cases and pulled 54% more relevant passages from the right court opinions.
Early access goes out through Trusted Access in ChatGPT and Codex, with API partners Harvey and Legora already building on top of it.
Take Away:
Legal research is exactly the kind of high-stakes, citation-heavy work where AI’s hallucination problem has burned lawyers before, so a dedicated, sourced index is a direct pitch for trust. If the accuracy numbers hold outside OpenAI’s own benchmarks, expect legal tech incumbents to feel the squeeze first.
Everything else in AI
Meta brought its Muse AI agent to Mac, letting users organize files, fill out forms, and pull info from connected apps without opening a browser.
Salesforce unveiled Koa, its first CRM reasoning model built on NVIDIA Nemotron, matching top models on CRM tasks with three times fewer errors.
Waymo announced plans to launch driverless robotaxis in Tokyo in 2027 with Nihon Kotsu and GO, its first commercial service outside the US.
Persona unveiled a $179 screenless AI wristband that lets users message, book appointments, and manage a calendar by voice alone, shipping in December.