expemilly

@LexnLin I thought Claude Design + gpt image was enough for ai design

@AnatoliKopadze Well, yeah - soon most businesses will be optimized for a system where 1-2 people control everything

this is f*cking gold my friend running a $40.000 MRR one-person business just shared his entire blueprint no team, no office, just Claude Code, a folder of skills and the right workflow the whole system in one image:

@maloymediika and does Obsidian connect directly to Claude?

EVERYONE TAKES NOTES WRONG. 1 AI TRICK MAKES THEM 10X SMARTER. Everyone chases a bigger model. Wrong question. The bottleneck is retrieval, not intelligence. A single router file inside Obsidian tells Claude Code where to look: wiki folder here, podcast notes there, daily logs below. It walks the map, pulls the right notes, answers with citations back to the source. Here is where it stings. People pay $20 a month for ChatGPT memory that forgets last week. A plain markdown folder plus one routing doc gives you a second brain that cites every claim and compounds every day. The audience sees a note-taking app. The operator sees a queryable memory layer.

@cyrilXBT yeah, oratory skills are insanely important for social success

A MIT professor gave a 1-hour lecture in 2019 that has 18 million views. He died 5 months after recording it. It was his final gift to the world. Patrick Winston taught at MIT for 50 years. The smartest engineers on earth sat in his classroom. And he spent his last lecture teaching them the one skill their degrees never covered. How to speak. 15 lessons that will change how you communicate forever: Never open with a joke. Your audience is not ready to laugh yet. Open with a promise of what they will know by the end. Your ideas are like your children. You are too close to them. What is obvious to you is invisible to everyone else. Explain the obvious. The 5-minute rule: the first 5 minutes of any talk determine whether people will listen for the next 55. Spend more time on your opening than anything else. Repeat your most important idea 3 times in 3 different ways. Once is never enough. Build a fence around your idea. Tell people what it is NOT before you tell them what it IS. Verbal punctuation. Pause. Let the idea land before moving to the next one. Ask questions nobody will answer. Then wait 7 seconds. The silence is not awkward. It is processing. Never read your slides. Your audience can read. They cannot listen and read simultaneously. Use the board not the slides. Writing forces you to slow down. Slowing down forces clarity. Inspire before you inform. Nobody learns from someone they are not inspired by. End with a contribution not a summary. Tell them what you gave them. Not what you said. Never say thank you at the end. It is weak. End with something that lands. Stories make ideas stick. Data makes ideas understood. You need both. In that order. The quality of your communication determines the quality of your ideas in the eyes of the world. Not the ideas themselves. Practice is not preparation. Practice IS the skill. Patrick Winston understood something most people spend their entire careers missing. Your ideas are only as powerful as your ability to transfer them into someone else's mind. You can be the smartest person in the room and be completely invisible. Or you can master communication and make average ideas feel like breakthroughs. He chose to spend his last lecture teaching this. Watch it tonight. Bookmark this first. Follow @cyrilXBT for more lessons from the people who built the future.

@zodchiii I think Anthropic might shut down such an abuse

nobody should know this trick tbh free Fable 5 will write its own operating manual if you ask. that .md then runs on any model, even GPT 5.6 after July 19 generating it costs real money 1.Prompt Fable 5: make an operating manual of your approach 2.New Project, manual goes in Project instructions 3.Run it on Opus 4.8 or feed it to GPT 5.6: the Fable playbook on whatever model is cheapest that day this is one piece of the full agent stack. the rest is in the article below 👇

@edwardluox did you vibe-code this somehow? or is this a new update on mac?

My Mac notch now runs Claude Code, Codex, Cursor, Grok🆕, Antigravity, OpenCode, and Pi side by side. Agent finishes → notch pops. Needs approval → notch pops. One click back to that exact session. usage, sounds, plan review, everything. no more switching between apps.

@nikitabier as I understand it, replies are the foundation of promotion on

Replies: +3.15% Original Posts: +1.8% Small Account Reach: +1.19% Time Spent: -0.13% Unregretted Time Spent: ∞

@maloymediika u too

@expemillyweb3 legend

Karpathy just built an agent stack that gets smarter every night he sleeps. started as a throwaway gist. blew up in a week. the reframe kills every "AI memory" startup: your notes are the codebase. claude is the programmer. you stop asking questions - you build a system that asks them for you. 3 loops run it: /> ingest: drop a paper. subagent splits it into atomic notes, wires each one to everything you already know. /> query: ask anything. it answers from your vault, in your words, citing your own pages - never hallucinating. /> prune: 3:14am every night, a second agent walks all 1,847 notes. kills stale claims. flags contradictions. no vector DB, no embeddings, no $200/mo memory app - just markdown and a loop that never sleeps. your $500 "second brain" course covered maybe 17% of this. I stopped taking notes this week because claude does it now. full setup below:

@GVausort ofc, I hope I helped you

@expemillyweb3 $500 "second brain" course covered maybe 17%, bookmarked

@biektive it's logical that building a system from 3 models is much more productive than from just 1

AI content is bad because the same model writes it and approves it. That is insane. A real workflow needs an adversary: > 1 agent creates > 1 agent rejects > 1 agent fixes The article shows the part people keep skipping: AI work needs conflict, not more prompts.

@biektive soon he will show the world his mind give him time

@expemillyweb3 lowk one of the most underrated ppl in the industry

@cyrilXBT Claude + Obsidian is a very powerful combo for working with ai

YOU CAN BUILD AN AI SECOND BRAIN IN 15 MINUTES. No coding experience. No $1000 course. Here is the entire setup. Step 1: Download Claude Desktop. Step 2: Download Obsidian. Step 3: Create a new vault and start dropping .MD files into it. Step 4: Tell Claude Code to connect to your vault using Karpathy's prompt: gist.github.com/karpathy/442a6… That is it. Your entire knowledge base becomes searchable, connectable, and queryable by the most powerful AI model on earth. Every note you have ever written. Every idea you have ever captured. Every resource you have ever saved. Claude reads all of it, finds connections you missed, and surfaces insights from your own thinking that you FORGOT you had. Most people use Claude as a search engine. The people building second brains use it as an INTELLIGENCE LAYER on top of everything they know. That gap is the gap between asking Google a question and having a research partner who has read everything you have ever written. Bookmark this. Build it tonight. Follow @cyrilXBT

@chopeaceus I didn't think ai girls were still relevant

MAYA TURNED A COMFYUI Z-IMAGE TURBO FACE INTO $3.2K THIS MONTH WITH SHORT REELS She did not start by copying a real person. She built a fictional adult AI persona around one reference face in ComfyUI, using Z-Image Turbo to lock the first look before making anything public. The important part was not making one perfect image. It was choosing a face shape, hair lane, styling, camera mood, and keeping that character recognizable across every post. After that, she moved the reference into WaveSpeedAI and used Nano Banana Pro for controlled edits. New poses, new outfits, new locations, same fictional identity. That gave her enough variation to make the account feel alive without rebuilding the model from zero every day. Then ComfyUI came back for the video layer. The cleanest images became short walking clips, stairs clips, mirror-style clips, and simple motion posts that could survive Reels and TikTok without turning explicit. The public funnel was simple: short clips -> profile visits -> paid Telegram or Fanvue page -> adult-only private photo drops. The money equation was simple too: 1,500 profile visits a week x 4% paid clicks x $15 access = about $900 gross before renewals, extra drops, and repeat buyers. Run that for a month and the dashboard can get close to $3.2K without needing a giant audience. The operational constraint is identity drift. If the face changes every few uploads, the audience stops following a character and starts judging random AI images. So the real asset is not the prompt. It is the reference face, the visual rules, and the repeatable posting loop. ComfyUI makes the character. WaveSpeedAI gives it daily range. Short video turns it into distribution. The paid page is just where the attention gets priced.

@maloymediika try connecting Claude to it he'll refine this system to perfection

SOMEONE TURNED HERMES AGENT INTO A PERSONAL LIBRARIAN THAT ARCHIVED 103 LINKS ACROSS 32 DAYS INTO 352 CONNECTED NODES. Every link you send it gets read, tagged, and filed with context. Months later it hands the thing back when you actually need it. 423 connections between saved posts, threads, and ideas. Most people bookmark on X and never see that link again. The feed eats it. The tag system forgets it. Wrong frame. The question is not where you store a link. The question is whether the link finds you back. The May 2026 view renders every save as a dot on a network, not a dead row in a list. Tap a node, land back on the original post. Here is where it stings. Notion, Raindrop, Pocket all sell you storage. This one sells you retrieval. The audience sees a bookmarks app. The operator sees a second brain that answers back. How much of your bookmarks tab have you opened this year?

@steipete how many hours a day do you work behind the ai?

That's about as many sessions as my Mac Studio can take.

@claudeai there's a feeling that the extension will continue until the end of July/August

We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.

@0x_kaize finally I can run my own Hermes for free

Free AI API KEY Grok 4.5 The service constantly provides access to new AIs that are gaining popularity, and here we go again - only this time, it's Grok 4.5. Here's the setup: 1/ Go to: zenmux.ai/invite/555LC2 2/ Sign up 3/ Go to Models tab 4/ Choose: xAI Grok 4.5 While everyone else is paying for APIs, you can get them completely FREE. But you'd better hurry - access will be blocked again soon. By the way, if you're looking for other models, I highly recommend this service, which I've been using since the day it launched.

@kocer_eth living the sweet life is a whole science

@expemillyweb3 autonomous agents running background loops while you sleep is the ultimate goal honestly

anthropic quietly admitted fable 5 lies about finished work. they buried the fix in their own model guide. 3 lines. paste into any autonomous prompt. hasn't lied since - but the real trick isn't in the prompt. it's in what you're doing right now, reading this. opus is a chatbot. fable 5 is a night shift worker. one answers in a minute. the other runs for hours while you sleep. simple formula: trigger + fable 5 + verification loop = money while you sleep. If you're interesting in it: the 3-line audit prompt is in the article below.

@GVausort isn't it easier to just rent a server yourself?

THE $60 AMD BC-250 EVERY HARDWARE HACKER IS QUIETLY BUYING FROM EBAY MINING LOTS HAS 16GB OF PS5 GRADE GDDR6 AND JUST KILLED $459 A MONTH IN AI SUBSCRIPTIONS IN A PRAGUE APARTMENT 00:37 the guy in the video holds up a black card he calls the amd bc two fifty and says these have 16 gigs of gddr6 ram and he's made 100 videos on them, the same card is showing up in every serious homelab ai setup right now miroslav kolar is 24, ex ethereum mining rig operator in prague, and pulled the $60 amd bc-250 from a defunct mining farm auction lot after the merge killed his rigs, the board itself is ps5 semi custom silicon repurposed for consumer use each bc-250 has 16gb of gddr6 unified between the apu and the shared system memory, that's twice the vram of the $249 jetson orin nano and it runs qwen 3.6 14b at 25 tokens per second right out of the box on ubuntu server miroslav cancelled chatgpt pro, claude code max, cursor, copilot and gemini the weekend the board arrived, kept only a $20 chatgpt plus, and the $60 hardware bill against the $459 monthly subscription stack paid for itself in four days he's already running three bc-250 boards in parallel across his workshop pointed at different open source models, total hardware spend $180 for what the article's map calls the frontier tier at $1,700 follow me, i track every mining leftover killing $200/mo ai subscriptions so you don't have to also read the article below, seriously worth it