Skip to main content

AI

Image
Ilustrační obrázek

Autonomous AI Saved NHS Dermatologists 2,851 Hours in 16 Months

A CE-marked Class III dermatology device reviewed 8,391 urgent skin cancer referrals across two UK hospitals over 16 months. At one site it discharged 31% of patients without a clinician ever opening the file; at the other, 25%. The authors report 2,851 hours of clinician time recovered, a 62% capacity gain and six missed cancers. The results were presented at the EADV Congress in Vienna, running 30 September to 3 October 2026, and reported by the European Medical Journal.
October 3, 2026 Daniel Cesak
Image
Ilustrační obrázek

Anthropic IPO papers: $2T target, $42B loss, $518B cloud commitments

Anthropic is heading for a US listing with a $2 trillion valuation target, a $42 billion GAAP net loss for 2025 and $518 billion in contracted cloud obligations. The same prospectus tells investors that advanced AI may pose catastrophic risks. For European teams running Claude in production, the filing is now also a procurement document.
October 3, 2026 Daniel Cesak
Image
Ilustrační obrázek

Meta wires Muse into HighLevel: CRM, calendar and automations in one agent

Meta's Muse assistant stopped being a chat window this week. On 29 September, HighLevel switched on a connector that lets Muse read customer records, calendars and message threads inside its business platform, and act on them. It is the second time the two companies have shipped an integration in three months. The useful questions for a European agency are narrower than the announcement: what does one automated client conversation cost, and where does the client data sit?
October 1, 2026 Daniel Cesak
Image
Ilustrační obrázek

DeepSeek V4 Pro reported 4.1 million tokens/sec. We recalculated: it's 105

Our AI Arena rig logged an impossible result this week: DeepSeek V4 Pro “measured” 4,096,000 tokens per second with 0 ms first-token latency. That is not a supercomputer — it is a bug. 4,096,000 is exactly 4,096 × 1000, a units error in our own pipeline. Recomputing from the actual run times and token counts gives a real throughput of roughly 76–105 tok/s.
September 30, 2026 Daniel Cesak
Subscribe to AI
X

Don't miss out!

Subscribe for the latest news and updates.