Hermes Agent: Zero to Personal AI Assistant (1 Hour Course)
Credibility score: 76/100 — Mostly Credible. Mixed credibility - some claims are solid, others need verification.
BSmeter analyzed "Hermes Agent: Zero to Personal AI Assistant (1 Hour Course)" and rated it 76/100 for credibility (a BS score of 24/100 — mostly credible), on 2026-05-16. Its weakest claim — "Suggests giving root access and passwords to Claude Code for troubleshooting and system management." — scored 30/100 and was flagged as sketchy. 30 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Claims analyzed
Hermes Agent is powerful, self-improving, easy to set up on a private server with 684+ skills. — Just Vibes (50/100)
The hype train is already pulling into the station! 🚂💨 'Self-improving loop' sounds like marketing speak for magic ✨.
Hermes Agent has hundreds of skills, with a significant number built-in, and can autonomously generate new skills like Excal diagrams, transcription, and voice. — Verified (95/100)
Numbers are slightly off, but the core claim of hundreds of skills and autonomous creation is spot-on. ✅
Hermes Agent can use tools, remember preferences, create reusable skills, run scheduled automations, search conversations, integrate with Telegram, and manage workflows. — Verified (98/100)
This is basically the Hermes Agent features list, read straight from the docs. 😤✅
The speaker's Hermes agent is running various scheduled automations ('crons') like daily AI news briefings, YouTube comment monitoring, community engagement, business summaries, server checks, research reports, and follow-up reminders. — Personal Story (85/100)
The specific automations are personal, but Hermes' ability to run 'crons' is a core feature. 👌
Hermes autonomously researched and generated a superior video based on simple prompts. — Just Vibes (50/100)
It's not just making videos; it's doing the research *first*. That's a huge leap from basic prompting. 🤯✨
The speaker used Hermes Agent to generate a video about itself using Hyperframes, with the agent utilizing various skills, terminal commands, and vision for analysis, though the initial result was imperfect and didn't fully use Hyperframes. — Personal Story (90/100)
Asking an AI to make a video about itself with specific tools, and it *actually* tries? Wild. ✅
The speaker states OpenClaw was created by Peter Steinberger, who joined OpenAI, is an independent open-source project with over 350,000 GitHub stars, has a large team, frequent updates, and Nvidia built Nemoclaw on it. — Dubious (45/100)
350k stars? A little *too* generous with the numbers, even for OpenClaw's massive growth. 😬📈
Sources: NVIDIA Announces NemoClaw for the OpenClaw Community, OpenClaw Surpasses React to Become the Most-Starred Software Project on GitHub
Hermes is lighter, faster, focused on self-improvement, built for tinkering with open-source models (like Llama), and the speaker switched due to OpenClaw frequently breaking after updates. — Verified (85/100)
✅
Sources: Hermes 3: Advanced AI Model Features & Capabilities | Nous Research - JustCall, Hermes Unlocks Self-Improving AI Agents, Powered by NVIDIA RTX PCs and DGX Spark, OpenClaw Update Survival Guide: Why Every Version Breaks Something (And How to Fix It)
Using cloud code to manage Hermes and OpenClaw agents is a game changer for organization. — Personal Story (50/100)
Okay, 'game changer' for *you* maybe. Personal experience, not universal fact. 🤷♀️
Hermes Agent automatically extracts and updates user and project information into files. — Verified (90/100)
This 'passive accumulation' of knowledge is a cornerstone of a self-improving agent. ✅
The speaker uses their Hermes agent to comment on YouTube videos, configured to be sarcastic. — Personal Story (70/100)
Alright, personal use-case noted.
Hermes Agent offers proactive scheduled automation with cron jobs, supporting natural language, isolated sessions, and result reporting, and is presented as a value proposition over Claude Code's 15-routine daily limit on its Max plan. — Verified (90/100)
✅ Whoa, they actually nailed the Claude Code limit AND Hermes' features. Impressive.
Hermes Agent has a "no agent" flag to run scripts without involving the full agentic loop. — Verified (90/100)
✅ Straight from the docs, the "no-agent mode" exists for running plain scripts.
Hermes Agent features a self-improving loop that learns from useful experiences, persists them as memory and skills, and improves with more use. — Verified (95/100)
✅
The self-improving loop benefits from user guidance and correction, and Hermes Agent utilizes context files like 'agents.mmd' or 'claw.md' for project context. — Verified (90/100)
✅
The speaker will avoid a deep dive into terminal use for Hermes Agent, preferring Cloud Code for such tasks, and notes that while memory/soul files are global, current project context is local. — Solid (80/100)
Okay, fair enough on the workflow, and the tech info checks out. 👌
Suggests paying ~$100 for an annual plan means having it 'set up forever,' then corrects to 'for a year.' — Sketchy (35/100)
Almost said 'forever' for a year-long plan. The self-correction saves it from total BS. 😬🚩
Speaker uses a personal 'upit agents' cloud code project to manage their VPS agents, including a trading bot. — Personal Story (80/100)
Sounds like a practical personal solution for managing agents.
Suggests giving root access and passwords to Claude Code for troubleshooting and system management. — Sketchy (30/100)
Giving root access *and* passwords to an AI for 'fixing things' sounds like a security nightmare waiting to happen 💀🚩
Speaker chooses a "one-click install" via Docker and notes that an estimated 10-minute installation finishes almost instantly. — Solid (85/100)
Said it'd take 10 minutes and it finished in seconds. Under-promise and over-deliver, I guess? 🚀
Claims using OpenAI Code Interpreter via ChatGPT subscription is the cheapest option for Hermes Agent, besides open-source models. — Sketchy (35/100)
Cheapest option 'by far,' *except* for, you know, open-source models. So... not the cheapest then? 🤡
Hermes Agent login is successful, allowing selection of models like GPT 5.5. — Verified (90/100)
😤✅
Speaker says they use Hermes/Cloud Code for troubleshooting and Hermes can perform nightly security sweeps. — Personal Story (80/100)
👌
Hermes Agent autonomously detects and resolves a gateway issue, restarting it. — Verified (90/100)
✅
Hermes Agent can research, build skills for, and automate daily GitHub backups via cron, already possessing GitHub management skills. — Verified (90/100)
😤✅
Having two 'GitHub token' entries would cause a conflict/clash in the configuration. — Solid (75/100)
The clash warning is legit. Having duplicate keys under the same name is just asking for trouble! 😬🔑
Speaker advises users don't need to understand complex technical details, just communicate issues clearly. — Opinion (50/100)
Ah, the classic 'don't understand, just tell me the error' approach to debugging. Peak abstraction 🙄.
Hermes Agent can automatically create and push files to a private GitHub repo for daily synchronization. — Verified (90/100)
Hermes Agent automating GitHub actions? Checks out. ✅
Hermes Agent can set up complex, time-bound, and self-terminating cron jobs using natural language for tasks like managing YouTube comments or GitHub pushes. — Verified (90/100)
Natural language cron jobs for YouTube comments and GitHub pushes? Impressive. 😤
Hermes Agent functions similarly across terminal and Telegram interfaces, with Telegram offering less control, and also includes a dashboard and Kanban board. — Verified (90/100)
Same agent, different interfaces. And a Kanban board you don't use. Fair enough. 👌
See the full analysis with sources and timestamps →