Steph

@karpathy does exactly the same thing !

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

@IvanLandabaso love this !

Harvey AI went from $0 to $200M+ ARR in 36 months. The growth playbook behind the fastest legal AI company ever built 🧵:

@scaling_shields cold

met a guy last week making $43,000/month sending cold emails to people who left 1-star reviews for his competitors not joking he scrapes trustpilot and g2 for every 1-star review in his niche finds the persons linkedin pulls their email sends them one message: "saw your review of [competitor]. we built [product] specifically because of complaints like yours. want to see it?" reply rate: 11.4% for context the average cold email reply rate is 0.3% his is 38x higher because rather than just emailing strangers, hes emailing people who are already pissed off and actively looking for an alternative they JUST took time out of their day to write a paragraph about how much they hate the thing you replace theyre literally the hottest lead on planet earth and theyre free public information sitting on review sites he showed me his process: step 1: filter trustpilot for 1-star and 2-star reviews of top 3 competitors posted in last 90 days step 2: copy the reviewers name into linkedin sales nav step 3: pull email with apollo or instantly step 4: send the email within 72 hours of them posting the review while theyre still mad step 5: appointment setter calls every positive reply within 5 minutes close rate on these calls: 41% because the entire sales conversation is "what did they fuck up" and "heres how we do it differently" literally no convincing or educating needed no overcoming objections about whether they need the category they already bought it once they already know they need it they just need it from someone who doesnt suck last month he scraped 290 negative reviews found emails for 203 of them got 23 positive replies booked 19 calls closed 8 deals at $4,200 average contract value $33,600 in revenue from reading complaints the craziest part three of his customers told him they were ABOUT to switch to a different competitor before he reached out he intercepted them mid-churn one guy said "i literally had the contract open in another tab when your email came in" timing is everything most people are cold emailing prospects who dont even know they have a problem yet this guy is emailing people who just spent 20 minutes writing a essay about their problem on the internet every 1-star review is someone announcing "i have budget, i have pain, and im shopping right now" your competitors are generating your lead list for you theyre spending $50K on ads to acquire a customer that customer hates them writes a review and you email them for free and close them in one call right now theres probably 200 people who reviewed your competitor in the last 60 days theyre all reachable theyre all in market and theyre all ignored because everyone thinks cold email means emailing cold people these people are on fire go put it out p.s. if you want us to setup a cold email system that books you 10-30 calls per month - DM me "COLD" (you ONLY pay for qualified calls actually booked onto your calendar)

Claude office 3D

crazy claude

command line just got an upgrade - commandline.lovable.app my personal os

@omooretweets interesting...

I uploaded 200 pages of raw court documents to NotebookLM. It created a true crime podcast that is better than 90% of what's out there now... And ends with the hosts debating the ethics of the genre 🤯

@kritikakodes motion

I am a Vibe coder, scare me with one word.🤔

best motion ai tool out there? tried remotion and animations are basics :(

@antonosika Awesome !

Introducing Lovable for more general tasks. Lovable has always been for building apps. Today it also becomes your data scientist, your business analyst, your deck builder, and your marketing assistant. This is a big step toward what Lovable is becoming: a general-purpose co-founder that can do anything. See examples below.

@felixrieseberg yay !

We're shipping a new feature in Claude Cowork as a research preview that I'm excited about: Dispatch! One persistent conversation with Claude that runs on your computer. Message it from your phone. Come back to finished work. To try it out, download Claude Desktop, then pair your phone.

claude seems to live in the future

@rayansadri Totally agree… Basically what I experienced too - its basically claude code with skills -_-

Tried it out. So the product scans stuff, then it shows Reddit tells me people post about my company, surfaces competitors I already knew about ages ago, wraps it in a dashboard, and wants me to scream “OMG we saved $60k. At this point I’m convinced half of startup hype is just polished demos. How’s this the “ultimate CMO” for me

@Remotion very cool ! Will use it for our next launch video !

Remotion now has Agent Skills - make videos just with Claude Code! $ npx skills add remotion-dev/skills This animation was created just by prompting 👇

@IshwarJha @zackbshapiro Interested to collaborate with HAQQ :) DM !

@zackbshapiro Here are 16 AI agents working together to offer everything customised for Indian Law system github.com/ishwarjha/lega…

im on a mission to enable the next (or first!) one-person billion-dollar company. tools like @openclaw have shown us what's possible . . . but it's still too much for most people, feels insecure and can be a total time suck that's why in 7 days im launching @joinsagexyz, your ai cofounder that gets shit done and runs your business 24/7 autonomously deploy a custom agent right from imessage ( no app needed) join the waitlist: joinsage.xyz

This is insane ! It kills Openclaw, Clawdbot or Moltbot

Lawyers everywhere are already using AI, but this index finally shows what that actually looks like in numbers. ​ haqq.ai/whitepaper/leg…

@realtalishaw @Lovable Looking forward !

6 months ago I quit my job @Lovable and people thought I was crazy and maybe I was… but in 8 days I’m doing something even crazier I’m launching my own startup, hold me accountable and come along for the ride it’s going to be a wild one