What I actually use
Every pick below is something I run daily while writing and testing the playbooks. If a tool stops earning its keep, it comes off this page. Nothing here is required to use the books.
⚙️ Hermes — the agent this site is about
The Hermes agent
The local-first agent all my playbooks are built around. If you're starting out, pair it with my free starter prompts and you'll be productive on day one.
🛠️ The build stack
The official docs
Every install command and config option, straight from the source. The books reference these constantly — worth bookmarking.
The community
The Nous Research Discord is where I take the questions the docs don't answer. Active, technical, and friendly to beginners.
Hosting & domain
This site runs as plain HTML on simple, cheap infrastructure. Boring on purpose — it will outlive every framework.
📚 Worth your money
My own books
The full playbooks: 58 Prompts for Power Users ($29), The Complete Beginner's Guide ($12), and the Complete Collection ($35) — instant download, free lifetime updates.
The lab behind Hermes
Nous Research's own site — their research feed and model releases are worth following if you want to understand where the agent is heading.
🖥️ The hardware I run it on
RTX 3060 12GB — the default pick
The 12 GB card my $12 beginner's guide assumes you have. Runs 8B models at long context and 27B-class at Q4 with KV-quant — see the VRAM planner.
RTX 4060 Ti / 5060 Ti 16GB — the headroom pick
16 GB is the sweet spot for agentic work: 27B models at real context without the VRAM anxiety. Worth it if you skip steps.
RTX 3090 24GB — used market king
Still the cheapest 24 GB you can get. Two of them is how the 70B-class experiments happen. Watch for mining-card history.
Affiliate disclosure: As an Amazon Associate, Axel Verne earns from qualifying purchases. Links marked as picks may also earn me a commission if you buy through them, at no extra cost to you. I only list tools I use myself. My books never require any product on this page, and no one paid to appear here.