This digest focuses on what actually touches a small or mid-sized company's day-to-day: tools, costs, vendor risk and platform changes.
Analysts are doing the math on how much revenue the massive AI infrastructure build-out must generate
Capital spending by the big cloud providers has reached a level that demands exceptionally fast revenue growth to pay off. Why it matters: if the boom cools, today's cheap model pricing and free tiers get more expensive — avoid locking yourself into a single vendor. technologyreview.com
Leaders of major AI labs are now openly discussing slowing the pace and prioritising safety
Anthropic's CEO called for a brake on model development, sparking debate across the industry. Why it matters: safety rhetoric usually precedes stricter usage terms and regulation — write down your AI usage rules and data flows now. arstechnica.com
Automated AI accounts are filling social platforms with low-quality posts and comments
A growing share of platform content now comes from agents rather than people. Why it matters: organic social reach is losing value fast for small firms — put the effort into email lists, referrals and your own channel instead. arstechnica.com
Musk dropped his claim against Apple but is continuing the court fight with OpenAI
The antitrust case is now aimed squarely at OpenAI alone. Why it matters: long legal battles can affect model availability and integrations — keep your AI layer swappable. arstechnica.com
Apple's new operating systems bring a reworked Siri and end support for Intel-era apps
This is the final macOS release that can still run old Intel software through Rosetta. Why it matters: if your office still runs legacy Intel apps or older Macs, budget the migration now. arstechnica.com
An agent model worked for nearly half an hour and returned finished routes with map files
It used open data to complete the task, but the steps it took stayed invisible to the user. Why it matters: long-running agent tasks are ready for real analytical work, yet the lack of auditability is a business risk you must cover in your process. simonwillison.net
The cost of writing code has collapsed and value is shifting to understanding the problem
What remains of the job is figuring out what people actually want and defining it precisely. Why it matters: when commissioning software, pay for business analysis and user experience, not for lines of code. simonwillison.net