running OpenWebUI in docker on a DigitalOcean droplet. This setup allows me to do use a connected interface for working on my home computer, work computer, and phone.

using API tokens @ venice.ai to get access to a number of models.

I use RAG Knowledbase (provided as part of OpenWebUI) & system prompting of the models to get agents that support a few different lines of thinking. This setup is steadily improving and I'm learning how to better use it. (Side note: I think if you have a small team and want to collaborate, this is a cool platform for it... they have #channels similar to irc/slack as a beta feature... looking forward to seeing that develop further).

In the last three weeks, I've used this setup to create a bunch of posts from notes that I've been keeping in my .org files and gotten some application architecture design and coding support from a dolphin-qwen-coder model.

It's been a bit of a surprise to notice how different the various models are & learning which ways to use them has been helpful. As a for-instance, I've noticed the very large models are a lot better at taking tons of details and turning that into smaller parts.

OpenWebUI offers Functions/Tools/Pipes and such which should be valuable when I figure out how to use them for stringing functionality together.

It's cost me about $4 for the past month of Tokens & about $15 for the hosting (I should use a smaller droplet and save money here.

If anybody has good recommendations for tutorials on improving my text prompts (I don't use any of the image models) I would greatly appreciate any recommendations. Thanks