1. Im not a traditional developer but I have found enjoyment in building with claude. I make apps and automation systems that monitor information, I learned by just... doing it. Claude, Python, whatever tool gets the job done. I have no computer science degree just curiosity and stubbornness to learn from the internet and its cool participants. 2. Honestly I joined here cause imm tired of the gatekeeping energy in tech spaces. People love telling you that you cant learn to build real things through vibe coding or AI assisted development. I Came here to be around people who actually encourage each other. 3. My main goal is trying to get better at understanding ML architecture, specifically how to use smaller specialized models for the heavy lifting and only call the big AI models when you really need them. Learning by building it, breaking it, and asking better questions each time. And I just want to learn from people way more experienced and technical than me.

A friend of ours @Joseph Fioramonti built a tool called Constellations (If you have a watch or attended the first ever afternoon tea session. You'll know who I'm talking about). It measures something most people and most companies get wrong: the gap between what you think you respond to and what you actually respond to. Take it here 👇(also I am NOT getting paid for this and this is not some sponsored thing. Joe does really cool work) https://gen.constellations.app/constellations/survey/d269cab5-coca-cola/skool 📋 How it works: You'll see a grid of Coca-Cola images across two pages. Drag the green (+) dots to the images that make you want a Coke right now. Drag the red (-) dots to the ones that don't. Hit submit. That's it. 🧠 Why this matters: Every day we interact with systems that run on words. Search engines, AI tools, prompts, interfaces. The words we use are becoming instructions. They're becoming code. But here's the problem. If someone asks you "what kind of marketing works on you?" you'll give an answer. And that answer will be mostly wrong. Because desire and language live in different places. You feel a response to an image before you can explain it. You scroll past something or stop on something before your brain catches up with a reason. Constellations measures that gap. The space between what you say you want and what you actually respond to. This is the same problem companies spend millions trying to solve. It's the same problem you'll run into when you build anything that depends on understanding what people actually care about. And it's the kind of thinking that separates people who build things that work from people who build things that look right on paper. Take the survey. @Joseph Fioramonti will compile the results And provide a report shortly! He is an expert in branding and psychology and can come up with some really amazing reports.

Quick version in case you missed it: instead of using RAG to re-chunk your sources every time you ask a question, you compile each source once into a persistent markdown wiki. The LLM extracts concepts, writes entity and concept pages, updates cross-references, flags contradictions, and maintains the whole thing. Future queries read the pre-synthesized wiki. The part that clicked for me: the reason most of us abandon our second brains is that backlink and cross-reference upkeep is boring. The LLM doesn't care. It's happy to touch fifteen pages in one pass. I spent a couple of weeks turning Karpathy's pattern into a Claude Code plugin that actually scales (atomic pages, sharded indexes, BM25 fallback past ~300 pages). It also runs in Codex, Cursor, Gemini CLI, Pi, and OpenClaw through the skills CLI. Install in Claude Code: /plugin marketplace add praneybehl/llm-wiki-plugin /plugin install llm-wiki@llm-wiki Or in any other supported agent: npx skills add praneybehl/llm-wiki-plugin -a <your-agent> Five slash commands (init, ingest, query, lint, stats), stdlib-only Python, no dependencies. Plays well with Obsidian if you want the graph view. Repo: https://github.com/praneybehl/llm-wiki-plugin Karpathy's gist: https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f Curious if anyone here has tried the pattern themselves. What did you ingest first, and what broke before it worked?