Quick learning from running Hermes + Ollama on Windows

Sharing this in case it saves someone time.

I recently tried setting up a local AI agent workflow on a Windows PC using Hermes + Ollama. Straight away, I hit a key constraint:

- If you're running Hermes on Windows, you need to use WSL

- And importantly — Ollama also needs to run inside WSL, not on native Windows

After spending quite a bit of time getting everything configured, I ran into an unexpected limitation:

👉 When both Hermes and Ollama are running in WSL, Hermes doesn’t really behave like a full agent

Instead, it acts more like a front-end / interface to Ollama, rather than giving you:

- agent-style feedback loops

- personality or memory

- richer autonomous behaviours you’d expect from agent frameworks

So in practice, you lose a lot of what makes tools like Hermes (or alternatives like OpenClaw) compelling.

I later found a few other posts confirming the same behaviour, so this doesn’t seem to be just a setup issue.

TL;DR:

If you're expecting a full agent experience on Windows with Hermes + Ollama via WSL, it currently feels more like a UI wrapper than a true agent runtime.

Curious if anyone’s found a better setup or workaround on Windows.

1 comment

Shipping Skool

skool.com/shipping-skool

build and ship real software with AI, no coding needed. 7 live calls/week, starter kit, and a community that actually builds stuff

📞 UNLIMITED 1-1 Coaching

🤝 Leave A Testimonial

Leaderboard (30-day)

+23

+10