hemantv

Show BF: Codeblocks A flexible, privacy-first AI chat client for your desktop. Codeblocks v0.0.1 - TL;DR A flexible, privacy-first AI chat client for your desktop. Key highlights: Connect to anything - OpenAI, Anthropic, Google, xAI, OpenRouter, or local models (Ollama, LM Studio) Smart context - Long conversations that don't break, with auto-summarization and intelligent token management Total privacy - Everything stored locally on your machine as JSON files Powerful workflows - Fork conversations, regenerate responses, compare models, rich markdown editor Full customization - Per-chat model settings, system prompts, tags, themes, keyboard shortcuts Built for developers and power users who want control over their AI conversations, especially those experimenting with local LLMs. Current status: Early alpha (macOS focus, some rough edges)

https://drive.google.com/drive/folders/15Hf3N5LxvqT_hYjoc3gCEU4G3ftphSCF?usp=sharing

Hey HN,

We're often paralyzed by too many decisions, sometimes even small ones that don't matter much, yet we find ourselves stuck in endless loops of analysis paralysis.

At the same time, there are theories suggesting we don't really know where our decisions come from.

I built Jagora for myself to break free from small and mundane decisions.

In our current world of logic, this might seem like a crazy idea. The main goal I had was to give thought to questioning our decision-making process.

Solution:

Combine LLMs with ancient guidance tools like the I Ching and astrology. You can also chat with philosophers from older times if you prefer.

Jagora https://jagora.ai

Signup is required to prevent abuse.

Sample workflow: https://www.loom.com/share/9e493834f47f4517aac3a3f95ad525f9

Link to the tool: https://nihing-production.up.railway.app/workflows

So I had this crazy idea - what if we could do machine learning stuff right in the browser? No servers, no fancy setups, just pure browser power. And guess what? It frickin' works!

Here's the deal:

- Everything's happening in your browser. Like, EVERYTHING.

- Your data? It's chillin' in local storage. No sneaky uploads.

- Backend? What backend? We don't need no stinkin' backend!

Quick rundown of what I built:

1. Drag n' drop your data: Just yeet your CSV into the browser.

2. Make your data all fancy:

Turn text into numbers 'cause machines are picky eaters. See what your data looks like before you commit. No surprises! 3. Peep your data:

I threw in some graphs 'cause who doesn't love a good chart? Pro-tip: If it looks like a bell curve, you're golden. 4. Train that bad boy:

Pick what goes in, what comes out. Mess with some settings if you're feeling spicy. 5. See what stuck:

Find out which parts of your data are the real MVPs. 6. Extra goodies:

Wanna mash up some CSVs? I got you Why this is nice

- Cheap as free: Your computer does all the work. No cloud bills here!

- Privacy FTW: Your data never leaves your browser. It's like Vegas, but for numbers.

- ML for everyone: No more gatekeeping. If you can click a button, you can do ML now.

- Speed demon: Updates happen faster than you can say "machine learning”