Video: "Hermes Just Killed The #1 AI Agent" by Julian Goldie on YouTube.
What "killed" actually means in this context
It's YouTube language — the title is designed to get clicks. What Julian is actually arguing is that Hermes has now reached feature parity or better with leading commercial agents in the areas that matter most for business workflows: task persistence, multi-agent coordination, memory across sessions, and model flexibility.
That's a meaningful claim. A year ago Hermes was an interesting open-source project with a rougher experience than the paid alternatives. The gap has closed considerably, and in some respects — particularly cost and the ability to run locally — Hermes is now ahead.
Where Hermes now pulls ahead
The clearest advantage is the combination of persistent memory and the /goal command. Commercial agents tend to forget the context of a job between sessions. Hermes stores a model of your business, your preferences, and the history of work it's done — and uses that to stay useful over time rather than starting fresh every time you open it.
The /goal command locks the agent onto a target and keeps it working until the job is genuinely done. No re-prompting when it gets stuck. In practice, this matters a great deal for longer jobs — content campaigns, research tasks, site audits — where interrupting every few minutes defeats the purpose of having an agent.
The boss-and-sub-agent model also sets it apart. You can configure a lead agent that assigns work to specialist sub-agents running in parallel. One agent researches, one drafts, one reviews. Commercial agents that run sequentially just can't match the throughput.
What the multi-agent angle changes for a small business
Most businesses don't need to outperform a Fortune 500 content team. But they do benefit from the same basic principle: when an agent can divide a job and run parts of it at the same time, you get results faster. A three-agent swarm handling a keyword cluster — research, draft, review — genuinely gets more done in an hour than a single sequential agent gets done in a day.
In practice, multi-agent is most useful for recurring, structured work: weekly content, competitive monitoring, email handling. One-off tasks don't need a swarm. Setting up the swarm for routine jobs pays back quickly.
What's still missing from Hermes
Worth knowing: Hermes still requires a more involved initial setup than most commercial tools. The first hour is not plug-and-play. Model selection matters, and choosing the wrong model for a task produces noticeably worse results. And some of the most capable features — memory, skills, the HUD dashboard — need configuration before they do anything useful.
For a business owner who just wants something that works out of the box, the commercial alternatives are still less effort for the first few hours. Hermes is the better long-term choice once it's configured correctly.
Where this connects to NordSys
We install, configure, and integrate Hermes Agent for businesses that want AI agents doing real, repeatable work — not a demo that goes quiet after the first session. If you want an honest view of whether Hermes is the right fit for the specific jobs you need automating, that conversation is free and practical.
See our AI Agents service →