Notes from the operator
Practical writing about running managed AI agents — security baselines, backup design, the real cost of doing this yourself, and how we think about product decisions.
- Positioning··6 min read
Why single-tenant AI agents matter for businesses
Shared SaaS AI tools mix your business data with everyone else's. Single-tenant deployments draw a hard line — and they don't have to be hard to operate.
Read article - Security··8 min read
How we harden every AI agent deployment
An agent box is a privileged target: it talks to your messaging platforms, holds API keys, and can run commands. Here's what hardened means for us.
Read article - Operations··7 min read
Backups for AI agents: what to actually protect
Backing up an agent isn't 'tar /home and call it done.' The interesting parts are which directories matter, which ones must stay out, and how you actually restore.
Read article - Cost··7 min read
Managed vs self-hosted AI agents: an honest comparison
Self-hosting is genuinely affordable on paper. Here's what people underestimate about the real total cost — and when it still makes sense to do it yourself.
Read article - Tech··6 min read
What is Hermes Agent? An operator's introduction
Hermes Agent is the open-source runtime under every Superagent deployment. Here's what it is, what makes it interesting, and what running it well actually involves.
Read article - Product··5 min read
Where should your AI agent live? Telegram, Slack, Discord, or web
The integration decision matters more than people think. Each platform has different ergonomics, different audit trails, and different headaches. Here's how to choose.
Read article