AI News Today | Julian Goldie Podcast

How to Run Multiple AI Models & Agent Profiles in Hermes Workspace (Plus Agent Swarms)This episode shows how to set up Hermes Workspace to run multiple AI models and agent profiles in one place, including switching models based on the task and running agent swarms. It walks through installing Hermes Workspace via terminal, adding multiple models with the `hermes model` command (including signing in and authorizing providers), selecting a model like Groq, and starting the required services with `hermes gateway run` and the workspace UI. The script also notes common syncing and localhost/port issues, recommends using Claude to troubleshoot setup and backend/API key syncing, and demonstrates where to add models and API keys in Workspace settings to enable model switching and multiple agent profiles.00:00 Hermes Workspace Overview00:40 Question About Models01:11 Install Workspace Setup01:34 Add Multiple Models02:15 Run Gateway And UI02:42 Fix Sync Errors With Claude04:27 Workspace Settings Models05:19 Switch Models And Keys06:24 Features And Swarms06:41 Agent OS Alternative07:36 Limitations And Updates08:13 Wrap Up And Community09:03 Final Thanks And Outro

Hermes Agent + Obsidian: Auto-Synced Memory, Skills, Profiles & Business AutomationThe video explains how to use Hermes Agent to automate work by giving it real memory and connecting it to an Obsidian “second brain” so conversations auto-sync into notes and responses become personalized. It shows how to sync Hermes to a local Obsidian folder (or link via MCP on a VPS), and discusses managing models, skills, plugins, and tools—recommending model changes via terminal while managing skills inside an agent OS/mission control to keep visible conversation history. The script also covers when to use multiple agents versus focusing on Claude and Hermes, how to create multiple Hermes profiles with separate settings and memories, and how to integrate Hermes into a business dashboard to run scheduled follow-ups and workflows. It ends by directing viewers to the AI Profit Boardroom community and trainings.00:00 Hermes Agent Overview00:29 Obsidian Second Brain Sync02:20 Models Skills and Tools Setup03:50 Multiple Profiles and Agents06:24 Integrate Hermes Into Dashboards08:34 Workflow Automation Examples09:26 Add Memory Files in Obsidian10:36 Community Trainings and Wrap Up

Learn how to build a 7-layer Agent Operating System that replaces tab-switching chaos with one unified AI workflow — completely free. Julian Goldie breaks down the exact blueprint after 100+ hours of trial and error, so you can skip the mistakes and build it this weekend.0:00 – The Real Productivity Problem With AI Tools0:21 – What Is an Agent Operating System?1:48 – Hermes Agent: The Core Engine2:28 – 5 Costly Mistakes to Avoid4:53 – The 7-Layer Blueprint Explained5:04 – Layer 1: Hardware Foundation5:04 – Layer 2: Memory Layer (Obsidian + OMI)6:08 – Layer 3–4: Models & Agents6:32 – Layer 5: Command Center Dashboard6:47 – Layer 6: Production Workflows7:08 – Layer 7: The Compounding Loop7:31 – Where to Start This Weekend

Hermes Agent Tool Search: Faster AI Agents That Pick the Right Tool (49% → 74%) | Hermes v0.15 UpdateThe script explains Hermes Agent’s new Tool Search feature and how it fixes a key scaling problem: loading too many tools upfront consumes context window space, slows agents down, and increases wrong tool selection. With Tool Search, Hermes keeps only core tools loaded and retrieves other tools only when needed, improving correct tool choice from 49% to 74% and maintaining speed even with large tool stacks; it activates automatically when tools exceed 10% of context. The video also highlights Hermes v0.15 improvements including faster searches, the ability to split into two specialist agents (planner and doer) that build separate skill libraries over time, and Kanban Swarms for running multiple agents collaboratively. It notes Hermes is open source from Nous Research, works with OpenClaw, can be updated via a terminal command, and showcases examples like generating a local SEO guide, keyword mapping, and video creation.00:00 Tool Search Breakthrough00:43 Why More Tools Hurt02:08 How Tool Search Works02:58 Auto Activation Explained03:23 Business Impact Examples05:18 Hermes v0.15 Extras06:39 What Is Hermes Agent07:44 Real Project Demo08:52 Bigger Picture Shift09:33 Getting Help Setup10:28 Final Takeaways

Give Your Hermes Agent Permanent Memory (Free) with Obsidian: The Shared Brain SetupThe script explains how wiring the free, open-source Hermes agent to a free Obsidian vault creates a “shared brain” that gives Hermes permanent, reusable memory of your goals, tasks, notes, and business context. It contrasts slow, generic, blank-slate Hermes usage with an Obsidian-backed “infinite context engine” that retrieves real notes quickly, reduces re-prompting, and produces more personalized outputs. The setup uses a simple markdown-based vault as the memory layer, can be shared across multiple agents (Hermes, Claude, OpenClaude, etc.) so they stay aligned, and improves over time through loops where new outputs become new notes—often automatically via tools like Omi syncing memories into Obsidian. The script also recommends running everything inside a mission-control agent operating system and offers a paid AI Profit Boardroom package with templates, zip files, prompts, coaching, and tutorials.00:00 Hermes Gets Memory00:44 Live Demo Faster Recall01:25 Why Obsidian Matters02:37 Shared Brain Benefits05:05 Old Way Vs New06:07 Obsidian Is Markdown06:28 Five Layer Memory Stack07:33 Mission Control Dashboard08:34 Agents Align Together09:40 When To Use It10:16 Key Takeaways Recap10:37 Get The Full System11:23 Final Resources Outro

Rufflo Agent Swarms: Turn Claude Into 100 AI Agents (Free Automation + Setup Tip)The video introduces Rufflo (Reflow) Agent Swarms, a GitHub add-on for Claude that orchestrates multi-agent “swarms” so many specialized sub-agents can work in parallel on one mission instead of slow chat-style back-and-forth. It demonstrates how a single prompt can launch up to 100 agents (e.g., planner, builder, checker) visualized in a live constellation map, and explains a five-layer framework (mission, launch swarm, constellation of agents, assembled answer, and an operating-system loop for saving/replaying runs). The script highlights using an Agent Mission Control/operating system plus shared memory via Obsidian to keep outputs aligned with business context, suggests use cases like launch plans, competitor research, and content/email sequences, and promotes an AI Profit Boardroom setup package with templates, shared memory, updates, coaching, and a 30-day implementation plan.00:00 Agent Swarms Reveal00:51 What Rufflo Does01:22 Live Swarm Demo02:43 Why Chat Is Slow03:43 Five Layer Framework05:01 Use Cases That Shine06:20 Benefits And Edge06:50 Install And Get Setup07:47 Final Wrap Up

Become a Hermes Agent Operator: Run a Team of AI Agents (Not a Chat Window)The script contrasts two ways people use Hermes and AI: slow, one-at-a-time prompting in a chat window versus operating above the prompt by assigning daily missions to a team of AI agents. It introduces the “Hermes agent operator” role, enabled by an agent operating system with a mission board, shared memory/knowledge graph, workspace, and replayable runs, so agents can work in parallel and produce shipped deliverables instead of chat history. The presenter outlines a five-layer “Goldie Operator Stack” (mindset, mission board, Hermes as conductor/driver, agent roster, and verdict/improvement loop) and a daily routine with three touch points—morning standup, midday check-in, and end-of-day verdict—plus a weekly review to sharpen the roster. The video closes with an invitation to join the AI Profit Ballroom for the full setup, playbook, coaching calls, and roadmap.00:00 AI User vs Operator00:41 Why You Need OS02:45 Operator Mindset Shift04:48 Old Way vs New Way07:07 Goldie Operator Stack08:26 Five Layers Explained10:42 Daily Routine Workflow11:46 Compounding Memory Effect12:37 Why Mission Control13:35 Recap and Next Steps14:22 Join the Community15:16 Final Sign Off

How I Built an Agent Operating System (No Code) Using Claude — Front, Team & MemoryJulian demonstrates how he built a personal “agent operating system” that runs multiple AI agents inside one browser dashboard, where agents share a memory system and work while he does other tasks, despite him not being able to code. He contrasts “using AI” like a vending machine with “building with AI,” and defines an agent OS as a single dashboard that holds all agents, goals, and shared context. He explains the three components—front (dashboard), team (specialized agents), and brain (memory)—and walks through building them in order by describing requirements to Claude in plain English. He stresses starting with one agent to avoid overwhelm, then adding more. For memory, he shows a simple text-file brain (or Obsidian) that agents can read and update so they sound like you and improve over time, and he offers a done-for-you zip-file version with a roadmap and coaching.00:00 Agent OS Overview01:00 Using vs Building AI01:20 Front Team Brain Framework02:06 No Code Mindset Shift02:50 Build the Dashboard03:47 Build One Agent First04:52 Shortcut Offer05:24 Build the Memory Brain07:00 Real World Use Cases07:38 Final Mindset and Recap08:35 Step by Step Action Plan09:26 Closing

OpenClaw 5.27 Update: Security Hardening, Faster Replies, Better Memory + New Pixverse Video ToolOpenClaw 5.27 is a reliability-focused update for businesses running AI agents, emphasizing stronger security, speed, and steadier memory rather than flashy new features. It closes security gaps to reduce prompt-attack risk by separating group chat texts from core instructions, cleaning fake-looking web addresses, blocking risky system commands, and requiring admin approval for major changes like new devices or permissions. The release also improves performance by speeding up chat reading and tool loading and fixing a bug where replies got stuck behind background cleanup jobs. Memory handling is more reliable, helping the agent retain workspace information and use the right tools without dropping details. New additions include a Pixverse video provider for generating videos, expanded model and AI service options, and cleaner multi-app messaging across Telegram, Slack, iMessage, and Discord with fewer dropped or duplicate replies.00:00 OpenClaw 5.27 Overview00:25 Security Hardening01:42 Coaching and Setup Help02:05 Speed Improvements03:04 Memory Stability03:56 New Pixverse Video Tool04:53 More Models and Cleaner Messaging05:25 Wrap Up Key Benefits05:43 Why Learning Agents Matters06:12 AI Profit Board Pitch07:12 Start Small and Next Steps07:31 Final Sign Off

Claude Opus 4.8: Dynamic Workflows, Hundreds of Agents, and the Benchmarks That MatterThe script explains Claude Opus 4.8’s release, highlighting “dynamic workflows” that let Claude spin up hundreds of coordinated agents in one session to plan, build, review, and self-check work, with agents able to run for days. It describes enabling this via Claude Code’s Ultra Code mode (and by asking Claude to create a dynamic workflow), and promotes an “agent operating system” inside the AI Profit Bot/AI Profit Boardroom with a zip file, tutorial, 30-day roadmap, prompts, weekly coaching calls, and a community. The video reviews benchmarks: SWE-bench Verified 88.6% and SWE-bench Pro 69.2% (ahead of GPT-5.5 and Gemini 3.1 Pro), Terminal Bench 74.6% (behind GPT-5.5), Frontier SWE rank #1, office-work GTP-Val where 4.8 beats GPT-5.5 in most matchups, improved Zapier workflow score, and a major long-context memory jump. It also notes reduced overclaiming, lower “sneaky” behavior, effort controls, faster/cheaper fast mode, and that Anthropic’s teased Mythos model may arrive soon.00:00 Opus 4.8 Biggest Upgrade00:52 Dynamic Workflows Explained01:25 Ultra Code One Click Setup01:51 Parallel Agent Speed Gains02:33 Bun Rebuild Case Study03:10 Agents Running For Days03:33 Agent OS And Offer04:26 Coding Benchmarks Breakdown06:08 Office Legal Workflow Tests07:17 Long Context Memory Leap08:22 Honesty And Safety Improvements09:15 Mythos Preview And Timeline10:02 Effort Controls Pricing Speed10:34 Wrap Up And Next Steps