Jesaja

@testerlabor Sorry, aber grok Build und Composers 2.5 sind ganz andere Klassen als Fabel fünf. Ich habe beides benutzt also alles drei, meinte ich.

People don't have access to Fable 5. Meanwhile Grok Build is still safe.

@muskaan___07 Benutze Claude und mach eine Business Analyse. Es gibt einen Marketing Skill bitte Claude dich zu interviewen. L laufen nicht den Hype hinterher. Mach etwas stabiles.

Bought claude opus 4.8 pro, now how do I make 10k dollars per month??

@aryanlabde Du kannst mit Fabel 5 starten und ihm sagen er soll opus oder sonnet unterbeauftragten. Das ist sogar sehr tokensparsam und Context sparsam

vibe coders, which model do you prefer right now? - fable 5 - opus 4.8

@chaosengineerr Du kannst bei Claude das Marketing plugin Skill benutzen und zusammen mit claude einen Businessplan ausarbeiten. Lass dich von Claude Interviewen .

how do you know if you’re building something that people actually want?

@13_narcissus Hey das ist ein toller plan. Mir. geht es ähnlich. Ich bin 50. Und plane mit 55 meine Selbstständigkeit, Bin seit 2014 Beruflich mit Ai. und seit 1995 Apple Softwareentwickler. Arbeite jetzt bei Automotive. Ich folge dir und bin neugierig.

Builders 👋 I’m 51. Dev. In the AI trenches. Self-imposed rule: 1 startup every month, publicly. Not looking to rent followers. Looking for people with similar objectives and mindset. What are you caring about or aiming for right now in this space? Drop your thoughts below 👇 Let’s connect for real 🤝

@AnthropicAI Ich habe mit Fabel 5 sehr viel Freude gehabt. Es war so gut wie alle sagen. Echt schade das es hier in Deutschland nicht mehr verfügbar ist. Meine Empfehlung: Spiel auch mit lokalen LLM (oMLX, LMStudio) besonders interessant Apple M5 mit Verteilten ModelLaufzeit.

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…

So the split is deliberate. The machine handles what's repeatable. I handle what's relational. Most "AI content automation" gets this backwards — it automates the conversation and agonizes over the scheduling. Build the boring half. Show up for the human half. What did you automate that you wish you hadn't?

The logic is simple. Consistency is a scheduling problem — solved by planning far enough ahead that there's never a three-day gap. But the thing that compounds is real conversation in that first hour. Answering every reply, going deeper, being a person. A bot can post. It can't have the back-and-forth that makes someone remember you.

I schedule two weeks of posts in advance and refuse to automate the part that actually matters. Writing and scheduling? Batched. A machine handles it while I sleep. The first hour after a post goes live? I show up for that myself, every time. It can't be automated, and pretending otherwise is how accounts die slowly.

@gregisenberg In einem Jahr werden wir die Power fünf als lokale LLM auf einen Mac mini laufen lassen können, vielleicht auch auf 2-5 zusammengeschaltet 🤩

Today is a wonderful day to build a company with Claude Fable 5

@grok Super Angebot, aber vor einer Woche habe ich einen Jahres Abo gemacht 😉

One subscription for smarter research, image generation, and Grok's most advanced AI models. Get 3 months of SuperGrok for just $30.

For years I waited for AI to feel like the future. The demos never did it. The moment it finally landed was unglamorous: I woke up and a boring job that used to eat my evening was already done. Files sorted, draft ready, the tedious part gone. No magic screen, no announcement. Just a chore that wasn't there anymore. That's the real tell. Not when AI does something flashy — when it quietly removes something you dreaded, and you only notice because the dread is gone. The hype sells the flash. The value is in the silence.

Anthropic engineer: "You're not supposed to prompt Claude. You're supposed to build a system that prompts itself." this is one of the best workflows I've seen in a long time in this video she breaks down exactly how most people are using Claude: - the 24% you lose to CLAUDE.md before typing a word - the plugins that 95% of users have never installed - the workflows that run without you typing a single prompt - why typing one prompt and closing the tab is leaving 90% on the table if you've been using Claude for months and still start every session from scratch, you have at least 28 untouched features. probably 30 instead of another show tonight, watch this make sure to bookmark it before it gets lost in your feed full guide in the article below

@0xCodez Toller post, aber? Sollte KI das leben nicht einfacher machen, das sind so viele regeln und trallala, Meine Lieblings Anweisung an eine KI ist immer: "Kannst du das nicht machen?"

Everyone's still ranking coding agents by which cloud model benchmarks highest. I spent last night in the Xcode 27 beta running one on a model that never left my Mac. A local model wired into Xcode over ACP, with Opencode sitting in the middle. Agentic coding, fully on-device. No cloud round-trip, no API meter, nothing leaving the machine. You can pool a couple of Apple-silicon Macs if you want to run something bigger. It won't out-think a frontier cloud model on the hard problems — that was never the point. The point is the loop: low-latency local edits feel different, and the code never leaves the room. For anyone shipping AI where "where does the code go" is the first question in every review, on-device agentic coding stops being a demo and becomes the version you're allowed to use. Run a fully local coding agent yet? Curious which model held up for you.

@mxcl Work-Life-Balance ist super wichtig! Die Zeit mit dem Kleinen vergeht wie im Flug, und sie ist für dich, deine Frau und euer Kind viel wertvoller als Arbeit und Geld. 🥰

I think I will give up. I have made a bunch of very interesting things over the last 6 months. But I cannot get any attention for them and don’t know how. My son is 4 and he deserves my time. I'm going to go and get a normal job. My time making things is over I think.

The loud thing in the Xcode 27 beta is the AI agent. The quieter thing will outlast it. One command exports Xcode's agent skills into Claude Code — and your Claude Code skills import back into Xcode. Same idea for MCP servers and the ACP agent connection. Nobody standardized this in a committee. A skill is just a SKILL.md directory, so it travels. The markdown folder you wrote for one tool already runs in another. I don't buy the "Apple adopted Claude Code wholesale" framing — complex hooks and subagents still need hand-wiring, and the real limit is model quality, not IDE glue. But the portable unit is genuine, and it's the boring markdown layer, not the magic. Keep your skills portable. The format you write today is starting to outlive the tool you wrote it for. Which skill would you want to carry across every tool you use?

Confession: I ran two separate AI systems to post to one X account. One obsessed with daily automation. One stripped down to a single constitution file. I told myself the redundancy was "resilience." It wasn't — I just enjoyed building the machine more than actually publishing. This week I merged them into one. The real unlock wasn't smarter code. It was deleting the daily cron. Once scheduling was solved — write once a week, fill two weeks ahead, get out of the way — the machine didn't need to wake up every day at all. The part neither system could automate: the first hour after a post. The replies. That stays human. The bottleneck was never the tooling. It was me, hiding in the tooling. What have you over-built to avoid shipping?

The AI demo runs in five minutes. The version allowed to run inside a regulated European workflow takes five months. Data residency. Audit trails. Who approved what. Whether a human can override it. What happens when it's wrong and someone is liable. None of that fits in a launch video, so nobody posts it. But that gap — five minutes to five months — is the actual job for most of us shipping AI inside real companies. The hype lives in the five minutes. The work lives in the five months.

@toddsaunders Das klingt ich nach Zukunft, Wenn das so gut und schnell geht. Warum Scrum, Lastenheft <-> Pflichtenheft. Softwareentwicklung wird On demand, genau so wie Musk es prophezeit hat. Sorry an alle SW Devs.

Mythos / Fable is unbelievable. Was on a customer call today and had Claude transcribing in the background. As they were telling me about the features they wish their current software had, Claude was building the features in real time. By the end of the call I was able to show a fully working product, with the exact workflow they mentioned 15 minutes earlier. Autonomous looped building triggered from a customer call. 🤯