Imraan

@0xCodez i still don’t get what the difference between this and a ralph loop is

Anthropic research lead: "99% of our engineers are running swarms of 300+ self-improving agents. close the agent loop. Give the model a way to verify its own output" in a 20-minute session, Anthropic team member explains how to build a model that improves itself. Claude + loops + plan mode + dynamic workflows -that’s the secret. Watch the talk, then save the playbook below.

@nifinet this is great in theory but i doubt this runs converts in prod

@mattpocockuk running pi and neuralwatt

Folks who are running GLM 5.2, how are you doing it? What harness/provider are you using? Getting FOMO about an open weights model for the first time

@coreyganim I build a brain for all of our clients, typically, it's a hell of a lot messier than how you describe it. businesses' processes are messy, but they work, and most of the context lives inside someone's brain. agree with the principle, though, super valuable to have

There's an AI business model that hasn't blown up yet but will be HUGE within a year: "Second Brain as a Service" Businesses need their company context organized so AI can effectively use it. Here's the play: 1. Charge 2-5K to build the knowledge base 2. Load their existing data 3. Create the schema/rules for how it gets organized 4. Charge $500-$1K/mo to maintain it 5 industries that desperately need this: 1. Agencies Build a client delivery brain: -past proposals -call transcripts -client notes -campaign reports -SOPs Skill/plugin idea: "Client Context Retriever" that pulls the right client history before calls, reports, and strategy work. 2. Coaches/consultants Build a content + offer brain: -frameworks -client questions -sales calls -testimonials -old posts Skill/plugin idea: "Offer Angle Finder" that turns repeated customer language into post ideas, emails, and sales assets. 3. Local service businesses Build an operations brain: -quote templates -customer FAQs -technician notes -intake forms -reviews Skill/plugin idea: "Quote Prep Assistant" that drafts the first version of a quote using the customer's request + company rules. 4. Ecommerce brands Build a customer/research brain: -reviews -support tickets -product specs -ad comments -competitor notes Skill/plugin idea: "Customer Voice Miner" that finds recurring objections, product issues, and ad angles. 5. Real estate teams Build a deal/context brain: -property notes -buyer preferences -neighborhood research -past messages -follow-up history Skill/plugin idea: "Deal Brief Generator" that prepares the agent before every showing or follow-up. The entire business model is making a company searchable, usable, and easier to operate with AI.

if you're a business owner feeling the AI anxiety of how to implement it into your business, please read this first

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…

@SaidAitmbarek it was bro

thinking of ditching codex for claude max 20x just for fable-5 is it worth it guys?

@TestClipping @andrewmccalip i turned it on this morning, forgot about it and have been sat in terminals pretty much the whole day and Ive earned $15 running probably 4 terminals concurrently

@andrewmccalip @imraan Is it even really possible to earn 200$ a day or limit is just for the show?

Taking a few vacation days to focus on the absolute insanity around Kickbacks.ai. This started as another dumb weekend coding project and got WAY out of hand. I’m just a hardware guy who likes tinkering with weird new concepts, but this is starting to feel like OpenClaw 2.0. Going to speedrun the same loop: harden servers, fight fraud, onboard users, triage the flood of issues, cut deals, and try to keep the wheels on. Fight back the bot army, then focus on payouts. I’ll be live posting through it. DMs are open. Sorry in advance if replies are a little delayed. Wild 72 hours ahead.

@andrewmccalip ah im a terminal jockey 🏇 awesome project, enjoy the wild ride that will be the next 72 hours, good luck!

@imraan They are! In extension mode

@dessaigne a hybrid model is working very well for us. bill for the outcome yes, but also help the biz find what outcome they actually need (99% different from what they open with) and working back from there. ultimately every biz has the same core problems

A founder asked me how to build an AI-services business without it turning into a consultancy. My answer: the whole game is in how you charge.

@claudeai this

What are you building?

Claude Fable 5 has been out for a couple of days. Some projects people have already built with it:

@__paleologo im genuinely curious if the fable compute is proportional to the token price theyre charging

Clearly, Fable is doing a lot of work, and unleashing a ton of agents. To review a short technical note, it released 31 agents, coded simulations to verify my results, did "adversarial reviews". Eventually, it only made the assumptions slightly more rigorous. It is all good. For a four-page technical note+a little code, though, it consumed all my Pro session tokens, *plus* $17 worth of credits. It is ridiculously expensive. I have 20-page reports that are way more complex than this. I can see how Anthropic has entered the phase of market-clearing prices, yield management, and pre-IPO. I recall Boris Cherny saying in a podcast, "run Opus [4.6], not Sonnet. It's worth it". I feel comfortable saying that running your top-shelf model is *not* worth it anymore. Decreasing returns, on most tasks. Like in the real world, some people can be real smart, but real expensive.

@theo would pay for that tier yesterday

If Anthropic put out a $1000/month tier that gets 5x the $200 tier limits and also lets us keep Fable access, I'd do it in a heartbeat.

@CoinMarketCap what happens when an agent drains the wallet on a bad prompt or the user wakes up to 47 small trades they didnt approve

LATEST: 🤖 Coinbase launched "Coinbase for Agents," a platform letting AI agents connect to users' accounts to trade crypto and make payments autonomously.

@SuiNetwork agents doing millions of microtx a day still need identity, dispute resolution and someone to eat the failed tx costs

The agentic economy needs instant, near-free settlement. AI agents will need millions of microtransactions per day; legacy rails can’t keep up. Sui can.

@DiegoRamos25592 @ClaudeDevs shit bro I hadn’t thought of that, thanks

@imraan @ClaudeDevs Use codex when Claude reached the limit. BTW I never reach codex limit

We've reset 5-hour and weekly rate limits for all users. Enjoy Fable 5!

@stackzz theres not a chance your repo isnt complete slop with shipping 259 pr's overnight but i get the viral X attempt the games the game

259 PRs overnight is not a coding story. It is an operator story. Boris Cherny showed Claude Code workflows where /loop, nested subagents, and self-verification turned one prompt into hundreds of agents pushing roughly 40k lines of code. The loud take is “AI will replace engineers.” The useful take is much sharper: the new edge is knowing how to design the swarm, cap the blast radius, read the logs, and decide which output is allowed to survive. Prompting is getting commoditized. Operating the machine is becoming the scarce skill.