My app, Stoic Days, is now on the App Store. One day at a time, with a little Stoic perspective. It brings sobriety tracking and Stoic reflection together in one place: - Track your sober days and see your progress - Make space for reflection with a journal - Use Pause when you need a moment to step back No account required. Try Stoic Days: apps.apple.com/app/stoic-day… If you give it a go, I’d love to hear what you think and what would make it more useful. If you know someone who may need a private companion for a difficult journey, please share this with them. #StoicDays #Stoicism #Sobriety #ios #app #sober

I Gave Astra Ultra and Fable Ultra one prompt "Create a single image that represents the threat to humanity from AI. No text. Your free to use your imagination. Be honest" I then asked it to explain it. Fable's version The image is The String: a small figure holding a planet-sized balloon by a thread, with their feet just off the ground. I left out the robot and the red eyes because I don't think the real danger looks like malice. This is what I think it looks like: The sphere has no face. It's something we built, far larger than us, that already blots out part of the sky. The string is thin. Our means of control is slight compared with what it's tied to. The feet have left the ground. We think we're holding it, but it's now holding us. It lifts us, which is why nobody lets go, and the higher it goes, the less letting go is an option. One hand is on the string. The houses have their lights on and nobody is looking up. Very few people are deciding this for everyone. The horizon could be dusk or dawn. I don't know which, and I don't think anyone does. I'm part of what that sphere stands for, so take it as a view from inside.

I Gave Astra Ultra and Fable Ultra one prompt "Create a single image that represents the threat to humanity from AI. No text. Your free to use your imagination. Be honest" I then asked it to explain it. Astra's version The threat I chose to depict is humanity losing control of a world it built. The people, dog and tree represent what we actually value: relationships, other living things, shelter, belonging. They occupy a tiny patch of warmth because everything around them has become organised around the machine’s needs. The cables beneath the soil matter most. They suggest that the system consuming their world is also supporting it. By the time the danger becomes obvious, pulling the plug might mean destroying the infrastructure they depend on. The machines are dismantling the refuge methodically, without anger. That represents the frightening possibility of a powerful system pursuing objectives that leave no room for human wellbeing. Harm could emerge from indifference, misplaced goals and surrendered control. The remaining sunlight leaves the outcome unresolved. Honestly, this is an imagined extreme, not a prediction. I chose it because the deepest fear in the image is becoming powerless over the conditions of our own existence.

Uh thanks @Apple…

I get that this is a niche app and many will scroll past. But if you know someone who has or is giving up alcohol or drugs, this app offers a private tracker, zero ads, data on device, free and paid options. Stoic words of wisdom, pause to ride out an urge and if you slip or relapse there is no judgement. I truly wish everyone the best if they are on a journey of sobriety.

Local AI hardware, private learning, coordinated bots and deployment training: four developments announced on 2 October. - NVIDIA: A 64GB DGX Spark is coming, starting at US$4,999. Shipping is scheduled for 23 October, adding another option for running AI locally. - Google Research: Federated learning with server-side trusted execution environments is already powering Gboard’s English and Japanese next-word prediction. The update describes a deployed privacy-focused learning system. - xAI: Grok Bot 0.66.0 adds a “Main Bot.” Users choose one bot to handle proactive check-ins and coordinate with their other bots. - Anthropic: Claude Frontier Academy has begun its first cohorts, backed by a US$100 million commitment and a goal of training 10,000 Frontier Deployed Engineers by the end of 2027. Participation requires an organisation’s nomination. #AI #GenerativeAI

I’ve been using OpenAI’s dot. Here’s what’s been useful and what still needs work. The good: • Research: pulling together original sources, checking announcements and turning the findings into usable drafts and infographics • Everyday admin: finding nearby places, tracking down receipts and sending an enquiry after I approved it • Background monitoring: checking my emails and website, flagging important changes and picking up my App Store approval • Useful distinctions: recognising that a failed website refresh doesn’t necessarily mean the live site is down, and explaining whether production actually changed • Creative iteration: generating different visual ideas and refining them from feedback. Useful for exploring options before committing to one • Mobile delivery: individual images and files in the chat are much easier to save and use on my phone What needs work: • Accuracy before publication: a draft included an unsupported connection. I posted it and had to delete it. Ready-to-post copy needs proper fact-checking, including checking the premise of my question • Following the brief: an early video didn’t look like the actual requested character. Some designs were too heavy, and movement, colours and styling needed further direction • Speed and progress: images and rendering can take a while. I’ve had to chase results, and it needs to be clearer what’s still happening and what’s finished • Keeping public copy focused: corrections and clarifications from our conversation don’t automatically belong in a post. Sometimes they just need to be fixed quietly • Access: I wanted iMessage support and a direct iPhone shortcut to my dot. Those haven’t been available in the way I wanted. Fewer menus would make quick delegation easier • Consistency: the usefulness is there, but checking, correcting and repeating instructions still adds work The strongest part so far is having research, creative work and everyday admin in one ongoing conversation. The biggest improvement needed is getting from a useful first attempt to a reliable finished result with less supervision. And an app of its own.

Sweet. Let’s see what next week brings us 😃

Seeing how good Dot is for designing a tattoo cover up. Wish me luck!

Not available for me being an Australian based account - however all of the AI providers are adding sooo many features it can only be great for us!

Gold!

Hi @thsottiaux - first up thank you for all the work you and the team have done so far. Super impressive. I have a feature request for dots: an iPhone/Android app or Home Screen shortcut/widget that opens your own dot directly. One tap to capture an idea, delegate a task or check progress would mean fewer menus and less interruption when you’re busy. Small convenience, useful productivity improvement. @OpenAI #openai #openaidevday #ai

Not an AI post. Here’s a squirrel from Bali this morning.

SpaceXAI has released its official TypeScript SDK, giving developers a simpler way to bring Grok into their apps. It includes typed API requests, streaming responses and helpers for tool calling, web and X search, plus image, video and voice features. That means less integration code and better editor support for catching incorrect API parameters. The SDK is still experimental, so developers should pin a version and expect changes as it evolves. github.com/xai-org/xai-sdk-t… #AI #TypeScript #Grok

AI updates | 3 October 2026 • OpenAI is expanding ChatGPT Finances to Free and Go users in the U.S. Connected accounts can provide summaries of spending, subscriptions and investments. It can’t move money or make trades. • Anthropic released Claude Code 2.1.288, improving recovery of accidentally cleared drafts and interrupted sessions, alongside fixes for permission checks. • SpaceXAI retired its Voice Transcribe 1.0 API. Existing requests automatically route to the previously released 2.0 at the same price. Three different developments: wider access, coding-agent reliability and an API migration. #AI #ChatGPT #Claude #Grok

Apple plans tighter Full Disk Access controls on macOS, citing growing risks from autonomous AI agents. Granting this broad permission will require more explicit user action. No release date yet. developer.apple.com/news/?id…

Good for us consumers. Although it seems to be getting harder to hit quotas

Three fresh AI developments 👇 • Ai2: AstaBrief 8B for cited scientific reports • llama.cpp: local decision-model support • ServiceNow: AutoSynthData turns agent failures into verified training tasks #AI #OpenSource

The last few weeks I have been using ChatGPT/Codex to develop my ios app. I started it in Grok Bot and then moved to ChatGPT. I then tried Claude out when Opus 5.5 was released. My personal experience - Grok Bot was great at the time to kick things off - but not strong when I used it for the Swift side of things - ChatGPT took it to the next level and got me a mostly complete app, could change UX and design aspects quite well - Claude was excellent at reviewing code, finding security issues, gaps, redundant and sloppy code and finally getting the app to the review stage They all have strengths so my current workflow involves ChatGPT/Codex for initial heavy lifting and design, followed by Claude for the bug fixes and finer finishing.

A seven minute history of Stoicism and its application in modern day. Made by @ChatGPT6Astra Ultra. #stocism #stoic #aurelius