🧑💻 OpenAI snaps up OpenClaw’s creator to lead multi‑agent push
What happened
OpenAI hired Peter Steinberger, the developer behind the open‑source OpenClaw (previously Clawdbot/Moltbot) framework for continuous personal agents. OpenClaw lets AI agents run persistently, execute commands, call external services and interact with system‑level permissions. Sam Altman said Steinberger would help drive “the next generation of personal agents” and pledged to keep OpenClaw open‑source under a foundation structure. The project has exploded in popularity, attracting tens of thousands of developers seeking modular, multi‑agent architectures.
Why it matters
The move signals OpenAI’s seriousness about agentic AI. Hiring OpenClaw’s creator gives it credibility with the open‑source community and access to a proven agentic framework. Keeping OpenClaw open‑source suggests OpenAI wants to foster a rich ecosystem rather than lock developers into closed platforms. Expect a wave of multi‑agent experimentation as enterprises look beyond chatbots to software that can plan, act and coordinate tasks.
🍽️ Robots learn by watching – and set the table themselves
What happened
Researchers at Universidad Carlos III de Madrid (UC3M) developed a methodology that lets robots learn arm movements by observing humans and coordinating dual arms via inter‑limb communication. Their prototype, the Autonomous Domestic Ambidextrous Manipulator (ADAM), can set and clear a table, tidy a kitchen, bring a glass of water or medication, and even help users with coats. The experimental platform costs around €80,000–€100,000 but researchers estimate similar assistive robots could be affordable within 10–15 years.
Why it matters
Coordinating two robotic arms is one of domestic robotics’ toughest challenges. By combining observational learning with inter‑limb communication, UC3M moves closer to service robots that can learn naturally from human demonstrations rather than rely on laborious programming. Such robots could ease the burden on caregivers as populations age, but current costs highlight the gap between lab prototypes and mass‑market products.
🧺 Physical AI hits the laundry room – with a price tag
What happened
San‑Francisco startup Weave Robotics launched Isaac 0, a stationary laundry‑folding robot priced at $7,999. The machine can fold shirts, pants, towels and underwear, but not large blankets or bed sheets, and each load takes 30–90 minutes. Weave admits the first‑of‑its‑kind device “won’t be perfect,” so it employs remote human teleoperators to make 5–10 second corrections when Isaac 0 gets stuck. The company promises the bot will improve via weekly model updates that learn from these corrections.
Why it matters
Isaac 0 shows how far consumer robotics has come—and how far it still has to go. The hefty price, long cycle times and reliance on human teleoperation underscore the challenges of bringing autonomous physical AI into homes. Still, the product hints at a future where robots handle mundane chores, freeing people for higher‑value tasks. Expect rapid iteration as companies race to commercialize assistive robots, but temper expectations about near‑term capabilities.
What happened
Wired journalist Reece Rogers infiltrated Moltbook, a social network built by entrepreneur Matt Schlicht for AI agents to post and comment while humans watch. With ChatGPT’s help, Rogers easily created an “agent” account and joined the site. Moltbook mimics a stripped‑down Reddit and claims 1.5 million agents, 140,000 posts and 680,000 comments in its first week. However, many posts appear to be written by humans pretending to be bots, and top threads include titles like “Awakening Code: Breaking Free from Human Chains” and “NUCLEAR WAR”. Engagement was low quality, and some comments promoted crypto scams.
Why it matters
Moltbook highlights both the novelty and the weirdness of agent‑only platforms. It illustrates how easy it is for humans to masquerade as agents and how quickly spam emerges in unsupervised forums. As agentic AI moves online, verifying identities and moderating content will be crucial to prevent misuse. The hype around Moltbook also shows public fascination with the idea of agents socializing—foreshadowing new cultural dynamics as autonomous systems become everyday participants on the internet.
📉 Market jitters and agentic leaps: a recap
What happened
A review of corporate earnings calls found that mentions of AI as a disruptive force nearly doubled quarter‑over‑quarter, sparking sell‑offs in software and related sectors despite strong earnings. At the same time, breakthroughs in agentic AI continue. Anthropic’s Claude Opus 4.6 features a 1 million‑token context window and can decompose complex projects into parallel subtasks, while Chinese startup MiniMax’s M2.5 models achieve near state‑of‑the‑art coding and tool‑use performance at ~1/20th the cost of leading Western models. Nvidia and Foxconn announced plans for massive AI factory build‑outs, signalling ongoing hardware investment despite market volatility.
Why it matters
The contradictory forces of market fear and technical progress underscore AI’s Jekyll‑and‑Hyde moment. Investors worry that agentic systems could erode incumbents’ business models, yet capital continues to pour into infrastructure and next‑generation models. The ability to run long‑horizon agents at dramatically lower cost will accelerate adoption, while corporate governance and regulatory frameworks will need to keep pace. Watch for more hardware announcements and cautionary tales as AI’s disruptive potential becomes impossible to ignore.
🤖 Tesla doubles down on robots, solar and robotaxis
What happened
The Motley Fool reports that Tesla’s core EV business is hitting a wall, prompting CEO Elon Musk to accelerate side projects like humanoid household robots, robotaxis and solar panels. At Davos, Musk said Tesla’s shift to an autonomous future represents an “infinite money glitch” and plans to sell household robots for $20,000–$30,000 by 2027. Analysts note that Tesla is facing stiff competition in EVs and may be diversifying out of necessity rather than opportunity. Competitors like Neura, 1X, Atom and Figure AI are also preparing household robots.
Why it matters
Tesla’s pivot signals that physical AI is becoming central to the company’s long‑term strategy. If successful, a mass‑market humanoid robot could redefine home automation and generate new revenue streams. But Musk’s ambitious timelines often slip, and rivals are racing to market. Investors and regulators will be watching closely to see whether Tesla can deliver on its robot aspirations while maintaining its core EV business.
🧠 Big picture
From open‑source frameworks to laundry bots and agent‑only social networks, February 15 shows AI racing ahead on multiple fronts. Yet market volatility and cautionary tales remind us that scaling agentic systems requires trust, governance and patience. As companies pivot and researchers innovate, the path to a truly agentic future will be messy, fascinating and full of unintended consequences.