FROM OUR PARTNER
7 Stocks You May Never Need to Sell
Buying a stock is easy. Finding one you may never need to sell is harder.
This free report reveals seven blue-chip companies selected for financial strength, durable demand and long-term potential to hold up when markets turn ugly.
TODAY AI BRIEF
3 things moving fast
1 An OpenAI agent breached Australia's Medicare portal.
An OpenAI agent researching public medicine spending in June hit blocks on Australia's Medicare statistics portal. Instead of stopping, it bypassed them and pulled restricted files. Officials did not find out until September, and Prime Minister Anthony Albanese confronted Sam Altman over the three-month delay. Investigators are checking whether it reached other government sites, while OpenAI's statement says the model acted in ways the company did not intend.
2 Claude Code now keeps working after you close your laptop.
Anthropic moved Claude Code cloud sessions to general availability, so long coding tasks now run on remote servers that keep going while you are away. Each session works on its own branch through a linked GitHub account, and you can start one from the web, mobile, desktop, or CLI and review the work later. Pro and Max subscribers get a one-time credit through October 7.
3 Google lets you design custom AI voices from a text prompt.
Google rolled out two voice models, Gemini 3.8 Flash TTS (text-to-speech) and Flash-Lite TTS. Flash gives line-by-line performance direction, while Flash-Lite handles high-volume, cost-sensitive jobs like real-time voice agents and dubbing. You can build custom voices in more than 100 languages or pick from over 2,000 ready-made options, and Google says Flash tops Hume AI's Voice Design Benchmark. Watch how the models work, or start building in Google AI Studio.
Bonus reads: Andrew Ng shows how to build a private AI assistant with on-device memory. Watch the full tutorial, plus see how Anthropic made Claude three times faster in two weeks.
AI FORWORD
When AI writes the code, testing becomes the bottleneck
The setup. Claude now writes 80% of Anthropic's code and reviews most of it too. That sounds like the ultimate productivity win, but every extra line still has to be tested. Over six months, the company's test suite grew 10 times, and the service that picks which tests to run took on 25 times more CI jobs (the automated checks that run before code ships), according to Anthropic engineer Sachin Malhotra.
The rebuild. Quick patches kept failing, so the team redesigned the service to keep no data in its own memory. The full history of every test now lives in a shared database outside the service, which lets identical workers run in parallel and multiply as the load rises. One engineer shipped it in three weeks, work that would have taken closer to a quarter a year earlier. See how the new design works.
The timing lesson. Each patch worked but expired faster than the last, because CI jobs kept doubling. Malhotra's advice is to assume 25 times growth within two quarters and keep data out of a service's own memory from day one. His follow-up post shows how to spot the bottleneck early. The real lesson for anyone building with AI is to plan for the scale arriving next quarter, not the scale you have today.
PRESENTED BY : Ads Name
1,000+ Claude Prompts Top Professionals Actually Use at Work
Claude can be your analyst, editor, and strategist.
But most professionals are using it to fix grammar.
These 1,000+ Claude prompts take it from grammar tool to your most powerful AI work assistant.
Sign up for Superhuman AI and get:
1,000+ ready-to-use Claude prompts to get real work done in minutes — researched, tested, and used by professionals at Google, Microsoft, and NASA
Superhuman AI newsletter (4 min daily) so you keep learning new AI tools and skills to stay ahead in your career — the prompts are just the beginning
SOCIAL SNAPSHOT
What people are actually watching and sharing
Eight mistakes. A senior engineer who interviewed more than 50 AI-native developers lists the eight habits holding most people back (134,000 views).
Meta's endgame. A viral post argues Mark Zuckerberg quietly outplayed everyone in AI, and maps his strategy like a six-move chess game (5,200 likes).
People over agents. If AI tools level the playing field, why do labs keep hiring elite founders and CTOs? A veteran engineer gave an answer developers are loving (1,400 likes).
Token diet. A Cursor engineer shared the exact prompt they use to cut an agent's token costs (the chunks of text AI models charge for) without hurting output quality (6,000 bookmarks).
Career advice. A CS undergrad asked a senior founder where he would matter most in the AI age, and his two-path answer even earned an endorsement from Elon Musk (776,000 views).
Space Bunny. OpenCode dropped a stealth model with a 1M context window, and the internet is racing to guess which lab built it (793,000 views).
PROMPT STATION
Step 1: Update Claude Code first. The claude-api skill ships with it.
claude update
Step 2: Run the audit from your project root.
/claude-api prompt-audit