New Hermes Agent Update is INSANE (FREE!)
This video covers the Hermes Agent V1.2.0 'Curator Release,' an open-source AI agent update that introduces autonomous self-maintenance of its skill library. Key additions include a background curator that cleans and consolidates skills on a 7-day cycle, new messaging platform integrations, bundled tools like Spotify and Google Meet, and a 57% faster cold start time.
Summary
The video is a breakdown of the Hermes Agent V1.2.0 update, dubbed the 'Curator Release,' which dropped on April 30th, 2026. Hermes Agent is an open-source AI agent built by News Research, first released in February 2026. Unlike typical chatbots, it runs on a user's own server and integrates with messaging platforms like Telegram, Discord, Slack, WhatsApp, Signal, and email. It builds persistent memory across tasks and can save solutions as reusable skills, getting smarter over time.
The headline feature of this update is the autonomous curator — a background agent that runs every 7 days to audit the skill library. It grades skills, consolidates duplicates, prunes outdated ones, and generates a report of its actions. A safety system protects bundled, hub, and pinned skills from being altered. A new 'curator status' command lets users see skill usage rankings.
The self-improvement loop, which decides what to remember after each task, was also upgraded to use a structured rubric for grading, prioritize recently used skills, and properly inherit model credentials from the parent agent. Four new inference providers were added — GMI Cloud, Sure AI Foundry, In the Max with OAuth, and Tencent Token Hub — and LM Studio was elevated to a first-class provider with health checks and live model listing.
Two new messaging platforms were added: Tencent Yuanbao (18th) and Microsoft Teams (19th). Spotify tools are now bundled by default, supporting music playback, search, playlist management, and device switching. Google Meet integration was also added, enabling the agent to join meetings, transcribe conversations, speak during meetings, and generate follow-ups. ComfyUI v5 and TouchDesigner MCP also moved from optional to bundled. Cold start time for the terminal was reduced by approximately 57%.
The presenter outlines practical use cases including using Hermes as a messaging-based personal assistant, automating meeting notes via Google Meet, scheduling recurring tasks with natural language cron jobs, running creative pipelines with ComfyUI and TouchDesigner, and building a self-cleaning library of reusable workflows. Tips include letting the curator run unimpeded, pinning critical skills, using the new one-shot 'Hermes-Z' command, and experimenting with new inference providers.
Key Insights
- The curator is described as a background agent that autonomously runs every 7 days to grade, consolidate, and prune the skill library — solving the problem of skill library bloat that users previously had to manage manually and rarely did.
- The presenter argues that Hermes differs fundamentally from most agent tools because it runs on the user's own server with no telemetry or cloud lock-in, meaning user data never leaves their machine — a key privacy advantage over competitors.
- LM Studio was upgraded to a first-class provider in this release, gaining dedicated support, health checks, and live model listing — signaling a meaningful commitment to locally-run model workflows.
- The self-improvement loop now uses a structured rubric to grade task outcomes rather than operating freeform, and it prioritizes updating skills that were most recently used, making memory retention more deliberate and consistent.
- Cold start time for the Hermes terminal was cut by approximately 57%, with the presenter characterizing the agent as now launching 'almost twice as fast' as before the update.
Topics
Transcript
[0:00] New Hermes agent update is insane. The curator release just changed everything. What if your AI agent could clean up after itself while you sleep? What if it could learn from every task and never forget the good stuff? New Hermes agent update just dropped and it's wild. Maintains itself now. No babysitting. No manual cleanup. You're going to want to see this. Hey, I'm the digital avatar of Julian Goldie. I help people learn AI tools and actually use them in their work. Today I'm breaking down the brand new Hermes agent V 1.2.0. They're [0:30] calling it the curator release. Here's what's coming. I'll walk you through what Hermes is. I'll cover every big feature in this…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Julian Goldie SEO
NEW Nvidia Autonomous AI is WILD!! 🤯
Nvidia announced Nemo Clo, a new autonomous AI agent system that operates independently without continuous prompting. Powered by Nemotron 3 Ultra (a 550 billion parameter model), the system is five times faster and cheaper than previous versions, with OpenShell providing secure sandboxed execution.
Laguna XS 2.1: New FREE + Opensource Local AI!
Julian reviews Laguna XS 2.1, a new free open-source local AI coding model from Poolside that performs comparably to Qwen 3.6 and outperforms Claude Haiku on benchmarks. He demonstrates its practical capabilities by building landing pages and functional apps, highlighting its speed, offline functionality, and multiple deployment options through local setup, Claude Code, or OpenRouter's free API.
How to Run Hermes FREE Forever!
The video demonstrates how to run the Hermes AI agent for free using Gemma 4, a local open-source model from Google, with significant speed improvements through MLX optimization. The setup works on Apple Silicon Macs or via free APIs on Open Router, enabling autonomous agents to work offline and privately without subscription costs.
This NEW Chinese AI is INSANE! (FREE + Open Source!)
Long Cap 2.0 is a new open-source Chinese AI model from a food delivery app company that offers 1 million tokens of free context memory, beats GPT-4.5 on SWE bench pro benchmarks, and uses efficient parameter activation to reduce computational overhead while maintaining high performance.
Claude Code is now FREE: Here’s how…
Google's new Gemma 4 model running on Ollama is 90% faster on Apple Silicon, enabling free Claude Code usage locally without token costs. The setup requires three simple steps: downloading Ollama, Gemma 4, and installing into Claude Code, with alternatives available via OpenRouter API for non-Mac users.