← Hermes
Jack Roberts
Every Hermes Concept explained for Normal People
⏱ 41 min video · 4 min read29 May 2026Worth watching
TL;DR
A comprehensive beginner-to-advanced tutorial covering 21 core concepts of the Hermes AI agent, from basic setup to advanced features like parallel sub-agents, memory systems, and scheduled tasks. The video explains how Hermes differs from chatbots like ChatGPT and Claude by acting as a persistent, mobile, tool-equipped personal assistant that learns over time.
Key points
1
Hermes is an AI agent, not a chatbot — it has tools and takes real actions (booking flights, reading emails, controlling your computer) rather than just giving advice
2
Hermes runs on your own computer for free or on a VPS, and connects to 22+ messaging platforms (Telegram, WhatsApp, Discord) via one shared intelligence
3
You can power Hermes with any AI model via OAuth (free ChatGPT/Grok subscriptions) or API keys through OpenRouter, letting you switch between Claude Opus 4.7, GPT, DeepSeek, and others per task
4
Memory compounds over time using a memory.md file and SQLite full-text search across all sessions, and can be extended via Obsidian, Pinecone, and meeting tools like Granola
5
Advanced features include parallel sub-agents for team-style research, scheduled heartbeat/cron tasks for daily briefings, goal-tracking with turn budgets, and GitHub backup of your entire agent configuration
Actionable insights
→
Use OAuth with your existing $20 ChatGPT or Grok subscription for zero extra cost, and reserve Claude API credits for tasks requiring top reasoning like Opus 4.7
→
Follow the principle of least access — never paste API keys into chat, use terminal environment variables instead, and do not give the agent email-send permissions until it is reliable enough
→
Clear chat sessions regularly and use the /compress command to prevent context window bloat, which degrades performance and inflates token costs significantly
→
Set up a soul.md file covering your identity, communication style, values, and goals so Hermes behaves as a personalised assistant rather than a generic chatbot
→
Use /background and parallel sub-agents to run multiple research streams concurrently, cutting multi-hour tasks to a fraction of the time
Notable quotes
“Some guy actually spent 4 million tokens in 2 hours of light usage. 21,000 tokens just asking about the weather because he made a mistake.”
“The idea with Hermes is we can take it anywhere with us and it enables us to be fully mobile, fully custom, and just unlock some incredible capabilities.”
“To a hammer everything is an L. But sometimes we want to use different things like Claude for design and ChatGPT for reviewing code.”
Worth watching?
✅
Worth watching the full video?
Watch if you are actively setting up or evaluating Hermes — the live demos of OAuth connections, MCP integrations, sub-agents, and cron jobs are genuinely useful and would take hours to piece together from documentation alone.
Saved you some time? The creator still deserves a like.
Want this for your own channels?
Add the channels you follow. Every new video summarised and in your inbox the moment it drops. From £4/month.
Try it free