Picture this: it’s a Saturday afternoon, I have a tea, my laptop is open, and I want to show you something. Not because I’ve figured it out yet, but because it’s the thing I can’t stop thinking about.
A few months ago I read about OpenClaw and autonomous AI agents doing all kinds of impressive things on their own. My first reaction was “wow”. My second was “…but is that safe?” Now Hermes AI Agents keep popping up in my YouTube feed, more and more often. Is it the safer successor? Is it just another system with a different name? I don’t know yet, and that’s exactly why I want to find out.
The idea: weekend projects
This is the first post in what I hope becomes a small series. The format is simple:
- I write down what’s in my head: the questions, the curiosity, the doubts.
- I spend a weekend poking around.
- Whatever I learn becomes its own post.
I don’t promise answers to everything. If a question is still open on Monday, that’s fine, and it’s useful information too.
What I’m curious about
The basics
- What is Hermes, actually? What’s the real difference from OpenClaw? Are they interchangeable, or just two names for similar ideas?
- Is it safer? And what does “safe” even mean here: access to my files, my accounts, my API keys?
Getting it running
- How do I install it, both the desktop apps (Mac and Windows) and on a VPS (I’m thinking Hostinger)?
- How do I connect to the server via SSH in a way that’s secure, but doesn’t take all weekend to set up?
- What are my options for a sensible security baseline from the start?
Connecting models
- How do I plug in the big models (GPT and Claude) and can I use my existing subscriptions, or do I need API access?
- Can I add open-source models through a service like OpenRouter?
- Which model makes sense for which agent? What do the different setups cost?
Agents and workflows
- Can I build a work environment with pre-defined agents for different tasks?
- I saw a Kanban-style system where agents work through cards one after another. Who creates the cards? Does one agent plan and hand tasks to the others? I’d love to just watch that happen.
What’s in it for me?
- For my work at the university: creating course materials, Moodle quizzes and announcements, small emails to students, drafting exercise sheets.
- For learning: can an agent act as a learning coach for my own studies?
- Beyond that: I keep reading about agents as life coach, health companion, productivity booster. What’s real, and what’s hype?
What comes next
If the weekend goes well, I’ll write up the setup (install, SSH, security), the model question, and the Kanban experiment as separate posts. If it goes badly, I’ll write about that too.
I’m writing this for fellow academics who are curious, maybe a little skeptical, and want to know what this could mean for their own work. Consider this me pulling up a chair next to you: let’s find out together.
