AI News Today | Julian Goldie Podcast

Hermes Agent V0.12 just dropped — and it's a massive update.This free release (the "Curator Update") brings 1096 code changes, 360 bug fixes, and 213 community contributors. Here's what's new:🧹 Autonomous Curator — runs every 7 days, cleans up old skills, merges duplicates, and keeps your agent sharp automatically⚡ 57% faster startup times🎵 Spotify integration — play music, search songs, manage playlists through your agent📹 Google Meet plugin — your agent can join calls, transcribe, and follow up for you💬 19 messaging platforms now supported (including Microsoft Teams and Tencent YABA)🤖 LM Studio, Comfy UI, and Touch Designer support added🔄 Remote model catalog — new models appear automatically without needing to updateTo update, just run: hermes update00:00 Introduction to Hermes V0.1200:54 How to Update01:26 Autonomous Curator Feature02:58 Performance Improvements03:18 LM Studio Integration03:54 Spotify Integration04:13 Google Meet Plugin05:17 What Changed Overview06:08 Faster Startup Times06:32 Spotify and Google Meet06:51 New Chat Platforms07:06 Additional Features and Providers07:32 Comfy UI and Touch Designer07:46 Remote Model Catalog08:08 Community Contributions08:25 Setup Guide and Recap09:23 Community and Resources

ChatGPT Codex Is Now a Full AI Agent (20% Faster): Email, Browser, Apps, Automations & WebsitesThe script showcases a major ChatGPT/OpenAI Codex update that turns Codex from a coding tool into a full AI agent for everyday work, including controlling the computer and browser, opening apps, managing Gmail, connecting calendar/files, and running scheduled daily automations. The presenter demonstrates installing a Gmail plugin, summarizing inbox requests, drafting/sending an automated daily email update with approval controls (auto-review/full access), and running tasks in parallel while Codex builds a website and navigates it autonomously. They highlight a cleaner design, role-based onboarding, automation templates, and improved computer/browser performance that’s 20% faster. Additional examples include creating slides and spreadsheets, using local apps like Calculator, building games, and having Codex set up other AI tools for non-technical users. The video ends by promoting a paid community and training with tutorials, coaching calls, and support.00:00 Codex Agent Reveal00:50 What’s New in Codex01:45 Connecting Gmail and Permissions02:41 Inbox Summaries and Scheduled Emails04:12 Parallel Agents and Website Build05:50 Browser Control Demo06:43 Slides Spreadsheets and Projects07:59 Local App Control and Reliability09:05 Automation Templates and Why It Matters10:18 Real World Use Cases11:23 Roadmap and Community Offer12:18 Final Thanks

OpenClaw 4.27 Release: Deep Infra Provider, Codex Computer Use, and Major Reliability FixesOpenClaw 4.27 focuses on reliability and stability, adding Deep Infra as a built-in provider for cheaper access to many AI models via one API key across chat, images, audio, video, embeddings, and more. It introduces Codex Computer Use so agents can control the desktop through a setup flow with fail-closed safety checks. File attachments in chat now properly stage non-image files (PDFs, docs, spreadsheets) instead of silently dropping them. Model selection is stricter, preventing silent model switching and making cron jobs fail closed when the chosen model is unavailable unless explicit fallback lists are configured. Major channel fixes improve Telegram (clear token errors, timeouts, preserved images), Slack (ping/pong keepalives, download timeouts, reset phrase handling), Discord (private-by-default replies and async interactions), and Matrix (no duplicate messages), plus faster gateway startup, outbound proxy routing, improved Windows restart handoffs, cleaner session resets, and more reliable memory “dreaming” with caps and retry behavior.00:00 OpenClaw 4.27 Overview00:36 Deep Infra Provider01:48 Codex Computer Use02:52 File Attachments Fixed03:30 Strict Model Selection04:44 Coaching Community Plug05:23 Telegram Stability Fixes06:27 Slack Reliability Updates07:35 Discord Threading Fixes08:37 Gateway Startup and Proxy09:58 Sessions and Memory Reliability11:09 Update Advice and Wrap Up

Hermes AI Agent + ComfyUI: Generate Free Local Images, Video & Audio (No Subscriptions)Hermes’s AI agent has a new ComfyUI skill that lets you generate photorealistic images, video, and audio for free from your computer with no subscriptions, API costs, or complicated setup, and it can even be triggered from your phone. The episode walks through updating Hermes via the terminal, loading the ComfyUI skill, and choosing between a local install or a cloud option using Comfy Cloud (which requires an API key). It explains what ComfyUI is, shows how Hermes installs and runs it locally, and highlights model options including Flux (about 12GB) plus lightweight, general-purpose, and higher-quality video choices. The script also tours ComfyUI templates and utilities like image-to-video, segmentation, upscaling, and background removal, and ends by promoting the AI Profit Board community, coaching calls, tutorials, and courses.00:00 Hermes Unlocks ComfyUI00:55 Update Hermes to Install01:51 Load Skill and Choose Setup02:21 What ComfyUI Actually Is03:09 Local Models and Flux Options03:47 Templates and Editing Tools04:37 Flux Workflows From Your Phone05:22 Sidebar Tour and 11Labs06:46 Troubleshooting With Hermes07:11 Community and Training Outro

Set Up Hermes AI Agent in 1 Click (Free Cloud, Free Local, or Hosted in 20 Seconds)The script explains how to get the Hermes AI agent running quickly with minimal technical effort, outlining three one-click setup options: a free cloud model approach, a free local setup on your own computer, and a hosted “Max Hermes” option on the Minimax platform. It first instructs viewers to download and run Ollamaa, choose a model (suggesting Deep Seek for general tasks), then copy a provided command into the terminal to install Hermes. It contrasts cloud models (free but token-limited) with local models like Gemma four (lightweight but requires downloading and stronger hardware), and mentions recommended models and that Hermes works with over 70 skills. It also demonstrates launching Max Hermes in Minimax, noting tradeoffs like limited integrations, and ends by promoting additional Hermes and AI agent training in the AI Profit Board community.00:00 Hermes One Click Setup00:50 Install Ollamaa First01:08 Pick a Cloud Model02:02 Run Hermes Locally02:52 Cloud vs Local Tradeoffs03:20 Max Hermes Hosted Option04:16 Limits and Perks05:27 Recommended Models and Skills05:59 NVIDIA Nano Omni Option07:13 GitHub Help with Claude07:52 Wrap Up and Next Steps08:06 Training and Community Plug

Run the Hermes AI Agent in Aion UI (100% Free) — Multi‑Agent Dashboard Setup GuideThe video shows how to run the Hermes AI agent inside Aion UI for free instead of using a terminal, highlighting Aion UI as a free, open-source dashboard for managing multiple agents like Hermes, Claude Code, OpenClaw, Gemini CLI, and others. It walks through downloading Aion UI from GitHub releases on Mac/Windows/Linux, opening the app, and verifying Hermes is connected and responding. The script covers key features including switching models, searching conversations, attaching folders/documents for project context, installing agent skills via a marketplace, configuring MCP tools, speech-to-text and image models, setting permissions for agents like Claude Code, and scheduling/automating tasks across multiple agents while keeping them awake. It also demonstrates remote access options (including from a phone), a desktop “pet” overlay, and creating agent teams with a leader agent coordinating others, then closes by promoting the AI Profit Boardroom community for training and coaching.00:00 Hermes Plus Aon UI Intro01:04 Why Aon UI Beats Terminal01:55 Download and Install Aon UI02:32 UI Tour and Skills Marketplace03:47 Connect Hermes and Multi Agent Chats04:22 Project Folders and Context Mode06:24 Scheduled Tasks Across Agents07:29 Models Integrations and Voice08:41 Remote Access and Desktop Pet10:08 Agent Teams and Wrap Up10:58 Community Offer and Outro

Claw X: The Free Desktop App That Makes OpenClaw Agent Management Easy (Mac/Windows/Linux)The script introduces Claw X, a free desktop app for managing Open Claw AI agents, positioning it as a cleaner, more reliable alternative to the standard Open Claw gateway, which is described as messy and buggy (especially scheduled tasks, provider/key management, and chat history). It shows how to download Claw X via GitHub releases or build from source, and highlights use cases like a personal assistant, automated monitoring via scheduled agents, developer productivity, and workflow automation. The walkthrough covers the improved UI; easier provider/model and API key configuration with token usage stats; creating and managing multiple agents; managing skills via a marketplace and local folder access; reliable scheduled task management with status views; better chat history navigation; and simpler channel integrations like WhatsApp and Discord. It also mentions multi-agent routing/parallel work and briefly notes Aon UI as an alternative for multiple CLIs, then promotes the creator’s AI Profit Boardroom community for tutorials, coaching, and courses.00:00 Claw X Overview01:03 UI Comparison01:43 Download and Install02:18 Use Cases and Benefits03:26 Models and Providers04:15 Agent Management05:18 Skills and Scheduling06:52 Channels and Integrations07:40 Multi Agent Workflows08:42 Alternatives and Demo09:27 Community and Next Steps

Setup Hermes Web UI: The Ultimate AI Agent Dashboard Guide

Learn how to set up and use the Hermes Web UI to manage your autonomous AI agents with a beautiful, responsive interface. This tutorial covers installation, switching between models like Deepseek and Claude, and managing advanced skills, memory, and scheduled tasks.

00:00 - 00:00 - Intro

00:40 - Quick Start & Installation

01:38 - Connecting Models & Providers

03:04 - Hermes Agent vs. Claude Code

05:31 - Managing Memory & Profiles

06:23 - Skills & Scheduled Tasks

09:28 - Using the Terminal Dashboard

Hermes Desktop Tutorial: The Best Free AI Agent Manager

Discover how to simplify AI agent management with the new Hermes Desktop app, a free tool for running local models and switching providers easily. Learn why Hermes is outperforming OpenClaw and how to set up your own virtual 3D office for your agents.

00:00 - 00:00 - Intro to Hermes Desktop

00:44 - 00:44 - Easy Setup & Prerequisites

02:49 - 02:49 - Hermes vs. Claude Code & OpenClaw

03:56 - 03:56 - Virtual 3D Office & Profiles

05:30 - 05:30 - Connecting Telegram, Discord & Email

07:11 - 07:11 - Skills, Memory & Persona Management

08:56 - 08:56 - Growth Trends: Hermes vs. OpenClaw

10:48 - 10:48 - Summary & Final Verdict

Give Hermes Persistent Memory for Free: OMI + Obsidian Context Setup (Works Across AI Agents)The video shows how to improve Hermes’ context and persistent memory by connecting it to OMI and an Obsidian vault, creating an automatic, fast, and largely free system that works across local AI agents (including Hermes and Open Claw). OMI records conversations and, with enabled screen recording and microphone access, captures daily activity to auto-generate organized “memories,” which can be exported into Obsidian as a growing memories file (over 650 entries in a few weeks). Hermes (and other agents) can then be pointed to the Obsidian file path to read and use these notes for context without manual updates or training. The script also mentions optional OMI features like wearable syncing, limited on-screen Q&A messages, and exporting to other apps (calendar, email, local files, Claude, ChatGPT, Gemini), plus a paid community/tutorial offering.00:00 Upgrade Hermes Memory00:31 OMI and Obsidian Setup01:20 How OMI Captures Context02:10 Syncing to Obsidian Vault02:51 Link Vault to Hermes03:28 Extra OMI Features04:15 Use With OpenClaw04:59 More Integrations and Rewind05:27 Wrap Up and Next Steps