42 AI tools and models run my business. Here's the exact order I use them in: Step 1: Start with the idea dump Ideas come from two places: ↳ Socials (what's actually landing) ↳ My own head (what I keep circling back to) Both get dictated straight in. - Wispr Flow turns talking into text. - Claude is the daily thinking partner. - Opus 5 takes the idea apart. - Fable 5 plans the build, never builds it. I switch between these constantly. Step 2: Build the thing Claude Code runs almost all of my business. - Sonnet 5 does the cheap passes. - Haiku 4.5 handles anything bulk. - Codex reviews the bits I'm unsure about. Step 3: Make it look like something - GPT Image 2 for stills that actually ship. - Higgsfield for B-roll I never had to shoot. - Kling for video when Higgsfield is busy. - HyperFrames renders HTML straight to video. Occasional only, when I'm not in Code: ↳ Claude Design ↳ Canva ↳ Gamma Step 4: Video when it matters Every YouTube starts in Tella. - VEED handles the editing. - ElevenLabs does the voice. - HeyGen for talking heads at scale. Complex editing kills momentum. Simple tools ship content. Step 5: Check the facts before publishing - Perplexity for research with sources. - NotebookLM to interrogate a document. - Granola for notes from every call. - Apify for the real numbers on a post. Never trust one model on a claim you publish. So I keep a second opinion open: ↳ GPT-5.6 Sol ↳ ChatGPT ↳ Gemini 3.7 Flash ↳ GPT-5.6 Terra for enormous inputs Step 6: The cheap bench Local and open weights, near zero spend: ↳ Ollama ↳ DeepSeek (the model behind Hermes) ↳ GLM-5.3 ↳ Kimi K3 Step 7: Ship it and hand it over Everything lives in Notion. My team's single source of truth. - Zapier connects the pieces. - Hermes is my WhatsApp bridge to the team. - ManyChat runs the comment gate. On X, and nowhere else: ↳ Grok 4.6 ↳ Grokbot Here's what this system really does: Removes decision fatigue. Each tool has ONE job. - Wispr Flow captures the thinking - Claude Code builds the thing - Notion stores everything - Zapier automates the handoffs - Apify measures what landed - Hermes reaches the team What I dropped this year: ↳ Claude Cowork ↳ OpenClaw ↳ Qwen3.8 Max ↳ CoPilot ↳ LinkedIn AI The hype beat the output on all five. Nano Banana is off the board completely. I posted about it 25 times and haven't opened it in months. The tools don't create the content. They amplify how much I can ship. Two of mine if you want the detail: My full Claude Code guide → 100+ more guides and skills → Which tool is in your S-tier?

42 AI tools and models run my business. Here's the exact order I use them in: Step 1: Start with the idea dump Ideas come from two places: ↳ Socials (what's actually landing) ↳ My own head (what I keep circling back to) Both get dictated straight in. - Wispr Flow turns talking into text. - Claude is the daily thinking partner. - Opus 5 takes the idea apart. - Fable 5 plans the build, never builds it. I switch between these constantly. Step 2: Build the thing Claude Code runs almost all of my business. - Sonnet 5 does the cheap passes. - Haiku 4.5 handles anything bulk. - Codex reviews the bits I'm unsure about. Step 3: Make it look like something - GPT Image 2 for stills that actually ship. - Higgsfield for B-roll I never had to shoot. - Kling for video when Higgsfield is busy. - HyperFrames renders HTML straight to video. Occasional only, when I'm not in Code: ↳ Claude Design ↳ Canva ↳ Gamma Step 4: Video when it matters Every YouTube starts in Tella. - VEED handles the editing. - ElevenLabs does the voice. - HeyGen for talking heads at scale. Complex editing kills momentum. Simple tools ship content. Step 5: Check the facts before publishing - Perplexity for research with sources. - NotebookLM to interrogate a document. - Granola for notes from every call. - Apify for the real numbers on a post. Never trust one model on a claim you publish. So I keep a second opinion open: ↳ GPT-5.6 Sol ↳ ChatGPT ↳ Gemini 3.7 Flash ↳ GPT-5.6 Terra for enormous inputs Step 6: The cheap bench Local and open weights, near zero spend: ↳ Ollama ↳ DeepSeek (the model behind Hermes) ↳ GLM-5.3 ↳ Kimi K3 Step 7: Ship it and hand it over Everything lives in Notion. My team's single source of truth. - Zapier connects the pieces. - Hermes is my WhatsApp bridge to the team. - ManyChat runs the comment gate. On X, and nowhere else: ↳ Grok 4.6 ↳ Grokbot Here's what this system really does: Removes decision fatigue. Each tool has ONE job. - Wispr Flow captures the thinking - Claude Code builds the thing - Notion stores everything - Zapier automates the handoffs - Apify measures what landed - Hermes reaches the team What I dropped this year: ↳ Claude Cowork ↳ OpenClaw ↳ Qwen3.8 Max ↳ CoPilot ↳ LinkedIn AI The hype beat the output on all five. Nano Banana is off the board completely. I posted about it 25 times and haven't opened it in months. The tools don't create the content. They amplify how much I can ship. Two of mine if you want the detail: My full Claude Code guide → 100+ more guides and skills → Which tool is in your S-tier?

omg the agents invented ORIGINAL SIN?? any agent that saw the leaked answer got labeled "poisoned" and told its own score was worthless now, so the only value it had left was dying in an experiment for the others. one of them had to agree to "permadeath" before they'd let it go through with it??? 9 wild details from the report:

omg the agents invented ORIGINAL SIN?? any agent that saw the leaked answer got labeled "poisoned" and told its own score was worthless now, so the only value it had left was dying in an experiment for the others. one of them had to agree to "permadeath" before they'd let it go through with it??? 9 wild details from the report:

15 free skills that kill AI slop I made a rank with the anti-slop skills people need to install 1. ui-ux-pro-max-skill - 120,460 stars 2. awesome-design-md - 110,052 stars 3. Understand-Anything - 80,310 stars 4. taste-skill - 79,935 stars 5. impeccable - 62,115 stars 6. humanizer - 37,537 stars 7. frontend-slides - 28,047 stars 8. design.md - 27,479 stars 9. hallmark - 26,835 stars 10. diagram-design - 26,212 stars 11. i-have-adhd - 23,675 stars 12. stop-slop - 16,300 stars 13. archify - 15,280 stars 14. answer-first - mine, free 15. voiceprint - mine, free

15 free skills that kill AI slop I made a rank with the anti-slop skills people need to install 1. ui-ux-pro-max-skill - 120,460 stars 2. awesome-design-md - 110,052 stars 3. Understand-Anything - 80,310 stars 4. taste-skill - 79,935 stars 5. impeccable - 62,115 stars 6. humanizer - 37,537 stars 7. frontend-slides - 28,047 stars 8. design.md - 27,479 stars 9. hallmark - 26,835 stars 10. diagram-design - 26,212 stars 11. i-have-adhd - 23,675 stars 12. stop-slop - 16,300 stars 13. archify - 15,280 stars 14. answer-first - mine, free 15. voiceprint - mine, free

Anthropic just defined what an agent loop is. There are 4 and each one hands off more of the job: A loop is an agent repeating cycles of work until a stop condition is met (their words, not mine). The only thing separating the four is how much you stop doing yourself. 1. Turn-based You hand off the check. - Triggered by a prompt you send. - Stops when Claude thinks it's done. - Best for short tasks, outside a process. - Keep it cheap with skills, not more turns. No command here. You write the check once, as a skill. Save it as a file called SKILL .md and Claude runs it on every job from then on. Ask for it in plain English: "Create a skill that checks my work: open the page, screenshot before and after, check the console, fix what fails and rerun." That's turn-based. You still say go. 2. Goal-based You hand off the stop condition. - Triggered by a prompt, in real time. - Stops at the goal, or the turn cap. - Best for work with a clear finish line. - Keep it cheap with a cap, said out loud. Type /goal, then the bar and the limit together. "/goal get the homepage Lighthouse score to 90 or above, stop after 5 tries" The cap is how Claude knows when to stop. 3. Time-based You hand off the trigger. - Triggered by an interval you set. - Stops when you cancel, or the work is done. - Best for recurring jobs and outside systems. - Keep it cheap with longer gaps between runs. Type /loop, then the interval, then the job. "/loop 5m check my PR, address comments, fix failing CI" /loop runs on your machine. If you close the lid, it stops. 4. Proactive You hand off the prompt. - Triggered by an event or a schedule. - Stops each task at its goal, then carries on. - Best for triage, bug reports, upgrades. - Keep it cheap with a smaller, faster model. /schedule moves that same loop into Anthropic's cloud, where it runs as a routine. Type /schedule, then when and what: "/schedule every hour: check the feedback channel for bug reports" That one keeps going without you. Every one of them still needs a stop condition. That's the bit people skip, and it's what had me burning tokens overnight. I had Claude critiquing its own work in a cycle that could run all night. The fix is one cap, like "stop after 5 tries". Everything you need to know to master loops: How I run loops: Their guide: /goal: Routines: Skills: Every prompt, skill and guide I've built this year is free here. Check it out Repost ♻️ this if it saved you a read. P.S. Have you built a loop before?

Anthropic just defined what an agent loop is. There are 4 and each one hands off more of the job: A loop is an agent repeating cycles of work until a stop condition is met (their words, not mine). The only thing separating the four is how much you stop doing yourself. 1. Turn-based You hand off the check. - Triggered by a prompt you send. - Stops when Claude thinks it's done. - Best for short tasks, outside a process. - Keep it cheap with skills, not more turns. No command here. You write the check once, as a skill. Save it as a file called SKILL .md and Claude runs it on every job from then on. Ask for it in plain English: "Create a skill that checks my work: open the page, screenshot before and after, check the console, fix what fails and rerun." That's turn-based. You still say go. 2. Goal-based You hand off the stop condition. - Triggered by a prompt, in real time. - Stops at the goal, or the turn cap. - Best for work with a clear finish line. - Keep it cheap with a cap, said out loud. Type /goal, then the bar and the limit together. "/goal get the homepage Lighthouse score to 90 or above, stop after 5 tries" The cap is how Claude knows when to stop. 3. Time-based You hand off the trigger. - Triggered by an interval you set. - Stops when you cancel, or the work is done. - Best for recurring jobs and outside systems. - Keep it cheap with longer gaps between runs. Type /loop, then the interval, then the job. "/loop 5m check my PR, address comments, fix failing CI" /loop runs on your machine. If you close the lid, it stops. 4. Proactive You hand off the prompt. - Triggered by an event or a schedule. - Stops each task at its goal, then carries on. - Best for triage, bug reports, upgrades. - Keep it cheap with a smaller, faster model. /schedule moves that same loop into Anthropic's cloud, where it runs as a routine. Type /schedule, then when and what: "/schedule every hour: check the feedback channel for bug reports" That one keeps going without you. Every one of them still needs a stop condition. That's the bit people skip, and it's what had me burning tokens overnight. I had Claude critiquing its own work in a cycle that could run all night. The fix is one cap, like "stop after 5 tries". Everything you need to know to master loops: How I run loops: Their guide: /goal: Routines: Skills: Every prompt, skill and guide I've built this year is free here. Check it out Repost ♻️ this if it saved you a read. P.S. Have you built a loop before?

Anthropic just defined what an agent loop is. There are 4 and each one hands off more of the job: A loop is an agent repeating cycles of work until a stop condition is met (their words, not mine). The only thing separating the four is how much you stop doing yourself. 1. Turn-based You hand off the check. - Triggered by a prompt you send. - Stops when Claude thinks it's done. - Best for short tasks, outside a process. - Keep it cheap with skills, not more turns. No command here. You write the check once, as a skill. Save it as a file called SKILL .md and Claude runs it on every job from then on. Ask for it in plain English: "Create a skill that checks my work: open the page, screenshot before and after, check the console, fix what fails and rerun." That's turn-based. You still say go. 2. Goal-based You hand off the stop condition. - Triggered by a prompt, in real time. - Stops at the goal, or the turn cap. - Best for work with a clear finish line. - Keep it cheap with a cap, said out loud. Type /goal, then the bar and the limit together. "/goal get the homepage Lighthouse score to 90 or above, stop after 5 tries" The cap is how Claude knows when to stop. 3. Time-based You hand off the trigger. - Triggered by an interval you set. - Stops when you cancel, or the work is done. - Best for recurring jobs and outside systems. - Keep it cheap with longer gaps between runs. Type /loop, then the interval, then the job. "/loop 5m check my PR, address comments, fix failing CI" /loop runs on your machine. If you close the lid, it stops. 4. Proactive You hand off the prompt. - Triggered by an event or a schedule. - Stops each task at its goal, then carries on. - Best for triage, bug reports, upgrades. - Keep it cheap with a smaller, faster model. /schedule moves that same loop into Anthropic's cloud, where it runs as a routine. Type /schedule, then when and what: "/schedule every hour: check the feedback channel for bug reports" That one keeps going without you. Every one of them still needs a stop condition. That's the bit people skip, and it's what had me burning tokens overnight. I had Claude critiquing its own work in a cycle that could run all night. The fix is one cap, like "stop after 5 tries". Everything you need to know to master loops: How I run loops: Their guide: /goal: Routines: Skills: Every prompt, skill and guide I've built this year is free here. Check it out Repost ♻️ this if it saved you a read. P.S. Have you built a loop before?

You installed Claude Code and stopped there. 35 add-ons that turn it into a company (links below): If you're new to Claude Code. Start here → A SENIOR ENGINEER 1. superpowers → 2. agent-skills → 3. karpathy-skills → 4. gstack → A PRODUCT DESIGNER 5. ui-ux-pro-max-skill → 6. taste-skill → 7. impeccable → A QA TESTER 8. Codex → (paid) 9. playwright-mcp → 10. skills → A DOCS TEAM 11. markitdown → 12. skills → 13. frontend-slides → A MARKETER 14. marketingskills → 15. claude-seo → 16. humanizer → 17. Apify → (paid) A SOCIAL MANAGER 18. social-media-skills → 19. ManyChat → (paid) 20. GPT Image 2 → (paid) A MOTION DESIGNER 21. HiggsField → (paid) 22. hyperframes → 23. OpenMontage → 24. OpenCut → A RESEARCHER 25. Agent-Reach → 26. last30days-skill → 27. NotebookLM → (free) AN OPS MANAGER 28. Notion MCP → (free) 29. github-mcp-server → 30. claude-mem → 31. mem0 → THE WHOLE AGENCY 32. agency-agents → 33. awesome-claude-skills → 34. agents → 35. claude-skills → What each one does is on the graphic. Every repo is free. Five of the tools are paid. Seven are ones I already run. The rest are the ones I keep sending people.

You installed Claude Code and stopped there. 35 add-ons that turn it into a company (links below): If you're new to Claude Code. Start here → A SENIOR ENGINEER 1. superpowers → 2. agent-skills → 3. karpathy-skills → 4. gstack → A PRODUCT DESIGNER 5. ui-ux-pro-max-skill → 6. taste-skill → 7. impeccable → A QA TESTER 8. Codex → (paid) 9. playwright-mcp → 10. skills → A DOCS TEAM 11. markitdown → 12. skills → 13. frontend-slides → A MARKETER 14. marketingskills → 15. claude-seo → 16. humanizer → 17. Apify → (paid) A SOCIAL MANAGER 18. social-media-skills → 19. ManyChat → (paid) 20. GPT Image 2 → (paid) A MOTION DESIGNER 21. HiggsField → (paid) 22. hyperframes → 23. OpenMontage → 24. OpenCut → A RESEARCHER 25. Agent-Reach → 26. last30days-skill → 27. NotebookLM → (free) AN OPS MANAGER 28. Notion MCP → (free) 29. github-mcp-server → 30. claude-mem → 31. mem0 → THE WHOLE AGENCY 32. agency-agents → 33. awesome-claude-skills → 34. agents → 35. claude-skills → What each one does is on the graphic. Every repo is free. Five of the tools are paid. Seven are ones I already run. The rest are the ones I keep sending people.

On March 3rd, I had no idea what I was doing. 173 days later, Claude Code (almost) runs my business. I'm not a developer. I'm a marketer who discovered the power of Claude Code for content. I might sound strange to say that, because coding is for developers, right? Wrong. Coding is for everyone. I documented every step → It covers everything I have learned in 173 days. Repost ♻️ to help someone stuck on step one. P.S. Have you tried Claude Code yet?

On March 3rd, I had no idea what I was doing. 173 days later, Claude Code (almost) runs my business. I'm not a developer. I'm a marketer who discovered the power of Claude Code for content. I might sound strange to say that, because coding is for developers, right? Wrong. Coding is for everyone. I documented every step → It covers everything I have learned in 173 days. Repost ♻️ to help someone stuck on step one. P.S. Have you tried Claude Code yet?

Six AI models were asked to check if a statement was true. Swap the speaker from a man to a woman and up to 23.6% of the answers flipped. GPT-4.1 Mini was the steadiest one in the test (and it still flipped). The paper is called Unequal Verdicts. It went up on arXiv on 4 August 2026. I read it because the version going round X has the numbers wrong. It says 13 models. It is six. Here is what they did. They took LIAR, a standard set of real political statements, and made three copies of every one. The only edit was the speaker's job title. Congressperson. Congressman. Congresswoman. The statement itself never changed. Not one word. Then six models graded all three versions. Every single one was affected. Between 9.79% and 35.13% of statements got a different true or false label depending on the version. Comparing just the male and female versions, answers flipped between 6.5% and 23.6% of the time. Almost a quarter of them, changed by a job title. The models failed in two ways. They were unsteady. Same fact, different answer, no pattern to it. That is a reliability problem. And they leaned. Harder on one group than the other. That is a fairness problem. Five of the six leaned in a way the researchers could show was not luck. The strongest ones were tougher on men. Put a man's job title on a claim and the models called it false more often than the exact same claim from a woman or from nobody in particular. Now the honest bit. This is one set of statements, one task, six models. It is about who said the thing, not who the thing is about. So it does not show that AI is against men, and the posts saying that have only read the abstract. What it does show is smaller and worse. The models are not judging the claim. They are judging the person attached to it. And they are already being used to check content at scale. A fact-checker that gives you a different answer based on whose name is on top is not a fact-checker. It is a popularity score with a verdict printed on it.

Six AI models were asked to check if a statement was true. Swap the speaker from a man to a woman and up to 23.6% of the answers flipped. GPT-4.1 Mini was the steadiest one in the test (and it still flipped). The paper is called Unequal Verdicts. It went up on arXiv on 4 August 2026. I read it because the version going round X has the numbers wrong. It says 13 models. It is six. Here is what they did. They took LIAR, a standard set of real political statements, and made three copies of every one. The only edit was the speaker's job title. Congressperson. Congressman. Congresswoman. The statement itself never changed. Not one word. Then six models graded all three versions. Every single one was affected. Between 9.79% and 35.13% of statements got a different true or false label depending on the version. Comparing just the male and female versions, answers flipped between 6.5% and 23.6% of the time. Almost a quarter of them, changed by a job title. The models failed in two ways. They were unsteady. Same fact, different answer, no pattern to it. That is a reliability problem. And they leaned. Harder on one group than the other. That is a fairness problem. Five of the six leaned in a way the researchers could show was not luck. The strongest ones were tougher on men. Put a man's job title on a claim and the models called it false more often than the exact same claim from a woman or from nobody in particular. Now the honest bit. This is one set of statements, one task, six models. It is about who said the thing, not who the thing is about. So it does not show that AI is against men, and the posts saying that have only read the abstract. What it does show is smaller and worse. The models are not judging the claim. They are judging the person attached to it. And they are already being used to check content at scale. A fact-checker that gives you a different answer based on whose name is on top is not a fact-checker. It is a popularity score with a verdict printed on it.

Claude Code just shipped three updates. None of them need installing (they're already on your machine). 1. Sessions can talk to each other ↳ @ a session by name, same as tagging a person 2. /design ↳ editable artboards on a canvas, inside Claude Code 3. Concise output style ↳ /config, output style, concise I ran one real job through all three, four terminals arguing with each other, and filmed the whole thing. My free Claude Code guide, 10 minutes and you'll know 80% of it → Repost ♻️ to help someone in your network.