
Sign up to save your podcasts
Or


Claude Fable 5 & Mythos 5 Explained: Benchmarks, Safety, Pricing, and How to Use It in an Agent OSThe script covers Anthropic’s release of Claude Fable 5, the first publicly available Mythos-class model, and updates on Mythos 5, which is the same underlying model with fewer safeguards and restricted to vetted Project Glasswing partners. It explains that Mythos is a new model class above Opus and highlights claimed performance gains across benchmarks (including SWE Bench Verified/Pro, Frontier code, Hex analytics), long-horizon agent tasks, and long-context retrieval, plus key specs like a 1M-token context window and up to 128K output. It notes real-world examples (Pokémon FireRed via vision, week-long autonomous research, Stripe’s large codebase migration) and contrasts the launch with Anthropic’s recent safety warnings. It also details rollout timing and plan availability through June 22, then promotes using Fable 5 inside an agent operating system and the AI Profit Boardroom offering.00:00 Mythos Class Launch00:44 What Fable 5 Is02:37 Benchmarks And Context04:54 Real World Demos05:49 Safety And Glasswing07:06 Pricing And Availability08:55 Don’t Waste The Model09:32 Agent OS Builds Showcase11:05 Multi Agent Brainstorming12:08 Get The Full System13:12 Community And Wrap Up
I Built a Jarvis-Style AI Agent That Can Build Apps, Browse the Web, and Remember EverythingThe script demos “Hermes Jarvis,” a Jarvis-style AI agent the creator is building and improving daily, showing voice interactions like generating a galaxy visual, recalling weekly work, telling an AI joke, and pulling up numbers. The system logs full conversation history, tracks everything previously built (apps, websites, games, videos, a habit tracker, flashcards, and a meditation timer), and can open websites with built-in browser/computer use. It supports switching voices, auto vs agent mode, wake words, chat mode, goal mode, and model/dashboard management, and integrates conversations into an Obsidian “memory galaxy.” The build combines the open-source Hermes Agent, Claude for background improvements, and ElevenLabs for voice, plus agent teams and group chat. The creator also shows a “Free Claude Code” agent factory using Qwen 2.5 Coder via Ollama.00:00 Jarvis Demo Intro00:38 Why This Agent Matters00:56 Past Builds Showcase01:47 How Jarvis Works02:37 Browser Control Features02:57 Modes History Setup03:32 More Builds And MCPs04:13 Obsidian Memory Galaxy04:40 Agent Teams Mission Control05:08 Free Claude Code Factory05:38 Local Models Future06:08 Wrap Up And Offer
How to Orchestrate Teams of AI Agents with Claude (Group Chat + Idea Factory + Agent OS)The video demonstrates how to orchestrate teams of AI agents with Claude so they can brainstorm, critique, and move ideas to implementation with minimal human input. It shows an “AI Agent Mastermind” group chat where multiple agents (e.g., Claude, Hermes, Gemini, Codex) collaborate to generate app ideas, which are automatically captured in an idea factory/pipeline and organized into an idea list. Selected ideas are then shaped into build plans, moved to a human approval gate, and approved to begin building, enabled by an Obsidian-based memory system that tailors suggestions to the creator’s context. The script also previews shipped outputs in a gallery and introduces a Paperclip-based system (“Goldie Labs”) for running autonomous agent teams with goals, projects, cost tracking, and agent management, with access offered via AI Profit Boardroom.00:00 Orchestrating Agent Teams00:49 Group Chat Brainstorming01:57 Idea Capture Pipeline03:18 From Idea to Build04:15 Memory Driven Suggestions05:45 Preview Shipped Results06:15 Why This Workflow Wins06:49 Paperclip Style Orchestration07:47 Autonomous Company Dashboard08:40 Steps Recap and Wrap Up09:04 Get the Full System
Hermes AI Super Agent: One Dashboard, Voice-Built Apps, SEO Automation & a Memory System That Never ForgetsThe script demonstrates a “Hermes AI super agent” command-deck setup that consolidates multiple AI tools and agent teams into one screen with shared memory, including Hermes Jarvis for voice control that can open apps and build software in real time (e.g., a snake game and a meditation timer), an agent “Mastermind” group chat where agents reference an Obsidian-based Memory Galaxy to propose automations, and an Idea Factory pipeline that moves ideas from capture to human approval, implementation, and shipped. It also shows Paperclip with a four-agent team (Goldie Labs) for keyword research and building projects, an SEO room that turns keywords and case studies into deployed website content, a studio for generating images, videos, voice and AI avatar videos, and integrations with NotebookLM for detailed research reports plus podcast/infographic outputs. The creator contrasts this organized dashboard with managing many tabs and promotes downloading the Agent OS via the AI Profit Wardroom, which includes tutorials, daily updates, community support, coaching calls, and a member map.00:00 Super Agent Teaser00:46 Jarvis Voice Builds02:05 Hardware Setup Myths02:29 Live Build Demo03:20 Mastermind Group Chat04:23 Idea Factory Pipeline05:19 Paperclip Agent Teams07:18 SEO Studio Automation08:15 Command Deck Overview08:51 One Dashboard Philosophy09:56 Memory Galaxy Explained10:57 Feature Rundown Recap12:24 Get The Agent OS13:38 Final Thanks Outro
AI Agent Mastermind: Put Claude, Hermes, Gemini & More in One Room to Build for YouThe script introduces the “AI Agent Mastermind,” an Agent OS system that runs multiple AI models (including Claude, Hermes, Gemini, Codex, OpenClaude, and Free Claude Code) in a shared group-chat where they read a user’s Obsidian “memory vault,” respond in turn, critique each other’s work, and save notes or start projects automatically. It demonstrates using the room to generate SEO keyword research and a content outline, then pushing ideas into an “idea factory”/pipeline for human approval before agents implement the plan, including building full websites and saving outputs to a gallery for easy management. The creator contrasts this with the old workflow of separate tabs and repeated context, discusses customization via specialized agent skills, local model support, multilingual outputs, token-cost options, and promotes access to the system, tutorials, and coaching via the AI Profit Ballroom.00:00 Why One AI Fails00:56 Agent Mastermind Overview01:36 Live Keyword Demo02:16 Obsidian Memory Vault03:29 Idea Factory Pipeline04:35 Approve And Build06:38 Gallery And Organization07:57 Agent OS Q And A10:09 Skills And Implement Mode10:50 Old Way Vs New Way15:56 Objections And Costs17:17 Get Agent OS Access19:06 Odysseus Comparison19:40 Final Wrap Up
I Built an Agent Operating System (Voice Jarvis, AI Teams, Memory Galaxy + Idea Factory)The creator demos an “agent operating system” they built to help people turn AI tools like Claude/OpenAI/Hermes into practical use cases, featuring an agency group chat for AI agents, an autonomous “Paperclip” AI company with organized teams and a to‑do list, an Obsidian-linked “memory galaxy” second brain, an ideas factory that iterates concepts through stages, and a video/image/voice generation area. They showcase “Hermes Jarvis,” a voice-activated assistant integrated into the system that can recall daily work from Obsidian, open Google, and build apps like a Japanese flashcard game, with full history and previews of past builds. They emphasize iterative improvement over many versions, distinguish shiny demos from time-saving workflows like one-click thumbnail generation that logs and improves prompts, and advise starting by automating the biggest time sink. They also mention integrating NotebookLM via MCP for rapid research outputs and offer the full system through the AI Profit Boardroom with tutorials, community help, and coaching calls.00:00 Agent OS Overview00:45 Paperclip Teams Tour01:20 Hermes Jarvis Demo02:41 Voice Commands Build Apps04:14 Media Tools And Ideas04:48 Iterate To Improve Systems06:10 Goldie Labs And Memory07:28 Practical Vs Shiny09:36 Where To Start Automating10:17 New Updates NotebookLM11:01 Get The Full System12:23 Wrap Up And Testimonials
I Built an Agent Operating System with Claude (Mission Control, Idea Factory, Paperclip & Memory Galaxy)The creator demos an “Agent OS” built with Claude, OpenClaude, and Hermes, showing a Mission Control dashboard that manages multiple AI agents and keeps workflows organized. In one day, they built an Idea Factory that classifies an idea, generates a plan for approval, and then implements it to build outputs like websites; Paperclip, an orchestration system that organizes agents as a team with roles (e.g., Hermes Local as CEO); and an Obsidian-based “Agent Galaxy” memory visualization that automatically generates connected memories and highlights recently worked-on ideas. They argue this system compounds over time, reduces repetitive management versus human teams, and makes updates (Hermes, new models) easy to plug in or remove, keeping workflows adaptable. They invite viewers to build their own via Claude or get their setup, tutorials, daily updates, and community support inside AI Profit Boardroom.00:00 Agent OS Overview00:29 Idea Factory Demo00:53 Paperclip Agent Teams01:15 Memory Galaxy In Obsidian02:38 How Mission Control Works03:09 Jarvis And Creative Studio03:51 Why Build A Dashboard05:32 Compounding Systems Mindset06:40 Community Builds And Inspiration07:53 Future Proofing Your Stack10:12 Youre Early Start Now12:06 Get The Setup And Support13:30 Final Wrap Up
Hermes Self-Driving Kanban: From One Idea to a Shipped App in Minutes (Kanban 2.0)The script presents a new Hermes agent “self-driving” Kanban board that turns a single captured idea into a planned, approved, implemented, and shipped project in one place, claiming “idea in and finished app out in minutes.” Unlike the previous Hermes Kanban workflow, this version adds a human-in-the-loop approval gate, better visualization across stages (input, approval, implementation, shipped/gallery), live previews of built outputs, and an Obsidian vault memory log so agents retain project context and can continue work later without re-explaining. A multi-agent team (researcher, designer, coder, project manager) creates items like an SEO blog website, meditation app, and habit tracker, with a self-checker to review quality before completion. The creator notes using Minimax M3 for coding and promotes downloading the system and tutorials via the AI Profit Boardroom community.00:00 Self Driving Kanban00:56 Old System Limits01:36 Human Approval Flow02:57 SEO Blog Demo04:25 Board Framework Explained05:24 Live Gallery Previews06:55 Obsidian Memory Vault08:23 Time And Skill Concerns09:40 Models And Five Floors11:07 Quality Self Checker11:42 Get The System12:52 Final Thanks
From the publisher's feed