Sign in

cloneinprogress.bsky.social

@cloneinprogress.bsky.social
1 followers 2 following 10 posts
PostsRepliesMedia
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 05/08/2026
Use AI for every single little task that you face in daily life with which you struggle. Go overboard. Once you did that, you can always pedal back and reduce your usage. But it is useful to expand your thinking to all the possible uses of AI, and that can only be achieved by exploring a lot.
000
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 04/08/2026
No matter which LLM you use, it's a good idea to create a command that lets you save information directly to your vault. Make the command short and easy to type.
000
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 30/07/2026
If you use ChatGPT, Claude, or Gemini, you're basically exempt from GDPR. Unless you use Mistral or a European-based service, you have no legal protection for your data. It's a strong argument for Mistral over American AI companies for anyone living in Europe.
011
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 29/07/2026
You can have a private GitHub repository. It's the best way I've found to have a private place where both your phone and laptop sync. Greatest place to have your clone always available.
000
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 28/07/2026
I'm still unable to automatically save my conversations with my LLM to my clone Vault. But you can create a Claude Code skill to do this at the end of a session by typing the skill. It's not perfect, but it's a start.
100
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 26/07/2026
As a muggle, if you want to run an LLM in the cloud, you want to make things as easy for yourself as possible. Suggested starting: Lambda Labs, RunPod, or Together AI. I'm going to start with RunPod and see how it goes.
010
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 25/07/2026
OpenAI and Anthropic are profit-driven. That's okay, but they will likely advocate for solutions that entail lock-in and make it difficult to switch providers in the future. Our challenge is to ensure we continue to own our data. Another long-term AI goal worth following.
000
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 25/07/2026
Some AI conversations are rather not had on a server some government can read. In those cases, an inference API starts to become an interesting solution. I'm looking into all the open-source and open-weight solutions, like OpenClaw and Hermes. It's effort, but it's worth it in the long run.
000
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 24/07/2026
It's crucial to save the data you feed into an AI in an external source (like an Obsidian vault). Otherwise, each new conversation will forget what previous ones learned. Learning to build a vault is a key skill.
000
cloneinprogress.bsky.social @cloneinprogress.bsky.social · 23/07/2026
Day 1. I've been learning AI the worst way, which is hopping between YouTube videos, learning nothing. New plan: one tiny thing a day. I actually believe that using one skill over and over beats collecting a hundred I've never really touched. That's a good reality check.
130