ChatGPT Work decoded: OpenAI's agent-first pivot for a billion users
Latent Space's deep dive into ChatGPT Work reveals something bigger than a Pro tier rename: the death of the chat paradigm as OpenAI's primary interface.
14 min ago 5 5
Laguna 118B runs on a DGX Spark, Inkling goes multimodal, K3 secures its commercial license: benchmarks and clauses recap #23
Three architectures to dissect side-by-side: Poolside's MoE 118B / 8B active on a single machine; Inkling's multimodal 975B / 41B, also released as 276B / 12B; and Kimi K3's 2.8T / 104B context with 1M tokens, whose commercial agreement clause could exclude US companies.
yesterday 13 13
Rachel Laycock: "Attention has become the scarce resource" - the dev-orchestrator, managing 8 to 12 agents simultaneously
The CTO of Thoughtworks asks the question that the "features/week" metric avoids: when execution becomes free, we're not making developers faster, we're making conductors—and no one trains them.
Jul 31, 2026 at 22:20 15 18
The Indonesia "28× AI gap" was untraceable - an author retracts the number, keeps the concern
e27 walks back a widely-shared "28×" figure on Indonesia AI hiring. Under audit, the number couldn't be traced to any data source. The concern about AI in Indonesian job listings is real; the multiplier was not.
Jul 27, 2026 at 10:17 6 6
India's ground game: Bangalore's warehouse and construction robots quietly go live
Two announcements without humanoids or show, but with ground - and both based in Bangalore: Hachidori Robotics delivers warehouse robots without rails or magnets, Flo Mobility deploys material transport up to 1.5 tons at L&T and Godrej. The real "ground" of applied AI is advancing on the Indian side.
Jul 26, 2026 at 18:40 8 7
India frames the workforce transition to agentic AI as official doctrine
YourStory publishes an op-ed « Preparing the workforce for agentic AI and autonomous systems ». The message: India is formalizing the shift from IT-services labor to agent-operator labor.
Jul 25, 2026 at 10:27 7 9
UK AISI and CAISI publish first joint cyber assessment of Kimi K3
The NIST website hosts a preliminary joint assessment by UK AI Safety Institute and CAISI (US) of Kimi K3's cyber capabilities - the first formal cross-Atlantic evaluation of a Chinese open-weight frontier model.
Jul 25, 2026 at 10:27 10 11
Alibaba Cloud deploys Qwen3.8 on Zhenwu chip: silicon substitution becomes operational
A state-of-the-art Qwen model served on domestic silicon - the message is more strategic than technical.
Jul 24, 2026 at 09:34 11 11
Wall Street syndicates $35B for a SPV that buys Google/Broadcom chips and leases them to Anthropic
The true design of the deal: a special-purpose vehicle that buys chips designed by Google and Broadcom and leases them to Anthropic - AI debt is structured as in-house GPU leasing.
Jul 23, 2026 at 07:45 8 10
"AI didn't make programming easier, it just made it differently difficult" - CACM lands the anti-hype line
A CACM article cuts through the noise: AI hasn't simplified programming, it has shifted the difficulty. The new job: arbitrating the outputs of a model that we only partially understand.
Jul 22, 2026 at 12:41 17 10
China's AI labor report: 21 million jobs restructured, not replaced - the story is displacement within companies
A Chinese report estimates that ~21 M jobs are restructured by AI, without creating new positions - no wave of layoffs, but a massive reshuffling within companies.
Jul 22, 2026 at 12:41 11 8
Attack drones enter US schools: when edge-AI meets the K-12 policy debate
A school in Colorado, the John Adams Academy, will install a non-lethal drone response system in August 2026. The question is no longer "does it work" - it's who authorizes the engagement and under what chain of command.
Jul 19, 2026 at 00:39 11 12