Anthropic Announces "Claude Tag," an Always-On AI Teammate for Slack
On June 23, Anthropic announced Claude Tag, a persistent AI agent integrated directly into Slack workflows. Claude Tag continuously learns organizational context and internal knowledge from Slack conversations and shared files, functioning as an "always-there AI colleague" that automatically answers team members' questions.
Beyond productivity gains, the feature carries strong strategic weight for Anthropic: the accumulation of organizational context in the enterprise. Analysts read it as a play to raise switching costs away from competing models by having Claude build knowledge graphs inside companies.
Meta Launches New AI Smart Glasses Lineup Starting at $299
Meta has announced a new lineup of AI smart glasses, with the most affordable model starting at $299. Building on the success of the existing Ray-Ban Meta series, the strategy is to expand the market with a wider range of price points and designs.
The smart glasses market has long been held back by a high-priced, niche image, but the $299 price point dramatically strengthens its appeal to everyday consumers. The big question is whether wearables with integrated AI assistants, cameras, and voice features can become the next major platform after the smartphone.
Seven Chinese Companies Are Already Shipping H100/H200-Class AI Chips
News that seven Chinese semiconductor companies are already shipping AI chips with performance rivaling NVIDIA's H100/H200 class has been drawing attention. Many of these companies have completed IPOs within the past six months, a sign that China's AI chip industry is maturing rapidly despite U.S. export controls.
Products from these companies, including Huawei's Ascend series, are becoming practical options, particularly for inference workloads. As the U.S.–China tech divide deepens, it is worth noting that China's build-out of its own AI infrastructure is moving faster than anticipated.
Utah Begins Pilot of AI Physicians — Practicing Doctors Voice Concerns
Utah has kicked off a pilot program for an AI-powered diagnostic system. The AI listens to patients describe their symptoms and proposes diagnoses, and its usefulness in real clinical settings is now being tested.
Physician groups on the ground, however, have voiced concerns such as "the diagnostic accuracy of AI is insufficiently validated" and "responsibility for medical care becomes ambiguous." AI adoption in healthcare depends not only on technical feasibility but also on legal and ethical frameworks being put in place first. The outcome of Utah's experiment could influence policy decisions in other states and countries.
Netflix Publishes Research on Controllable AI Video Editing
Netflix's technical team has released the results of early-stage research pursuing a "more controllable approach" to AI-assisted video editing. The work focuses on exploring editing methods that can capture a filmmaker's intent in fine detail—something conventional generative AI video tools have struggled with.
In content production, a long-standing complaint has been that "AI can generate the footage, but you can't fine-tune it to match your vision." Netflix's research offers a practical approach to this controllability problem and could shape production workflows going forward.
Krea 2 Released on Hugging Face — A New Option for Image Generation Models
Krea 2, an AI image generation model, has been released on Hugging Face. The community rates it highly for balancing high-quality output with inference speed, and it is attracting attention as a model that also takes local execution into account.
The open-weights release offers an alternative to Stable Diffusion–based models. Image generation is seeing intensifying competition between commercial models (Midjourney, DALL-E, etc.) and open-source models, and Krea 2's arrival is significant in that it broadens the range of options.
---