I built my own knowledge system that is a second brain for my work and life. Three separate knowledge bases. One for strategy, planning and personal usage, one for emails and work, one for all kinds of building. All connected through a graph index with Graphify. How it basically works Each knowledge base has its own folder structure (raw data, references, live state, logs). Every file gets indexed into a graph that maps connections between them. A graph query surfaces related files instantly, so my agent doesn't have to guess where something lives. Important decisions get their own file with context and reasoning so I can retrace them later. The whole thing auto-commits to a private GitHub repo every 6 hours. This made my agent stop digging through files or losing context. With this system, I can ask it "what did I decide about X" or "show me everything related to Y" and get accurate answers. What do you think? Tips or things you would do differently?

Anyone else out there trying out Fable 5 while it’s in available in this promotional period? I’m trying to max out its usage for the next week. It’s wild how much more intuitive and capable it seems to be compared to other models. I had it clean up my budget app and now working on a personal operating system for my disorganized life and and lead generation and CRM for my day job. What do you all have it working on?

If you jumped on Fable 5 this week and watched your usage bar drop like a rock, you're not alone, and you're not doing anything wrong. I did the exact same thing. My first couple of sessions, I burned through a big chunk of my weekly limit before lunch. And here's the frustrating part: the model is genuinely incredible. It's the strongest one I've ever used. But it's powerful enough to spend your tokens like water if you don't tell it not to. So I did what 31 years of engineering trained me to do when something is worth understanding. I studied it. I watched Nate Herk's video, "6 Simple Rules That Change How Fable 5 Works." Then I went straight to the source and read Anthropic's own Fable 5 prompting doc front to back. Then I ran my own tests to see what actually moved the needle on cost, not what sounded good on paper. Two things cause almost all the waste. You leave the "effort" set high when the task doesn't need it, and you reach for Fable for everything when you should be reaching for it maybe one time in ten. Then I took Nate's six rules and my own research and built them into one thing you can actually use. Not a cheat sheet you forget to open. A Claude Code skill that lives in your setup and steers every Fable session for you, automatically. The results so far have been excellent. Same model, same plan, and I'm getting top-tier work out of Fable 5 without hitting my limits anymore. I'm not going to gate this. Right below this post you'll find two things: the whole skill, ready to paste in, and a plain-English breakdown of what every part does and why it saves you money. If I were sitting where you are right now, watching that usage bar and wondering if Fable is even worth it, here's exactly what I'd do. Grab the skill, drop it in, and start every real task by asking one question: "Does this actually need Fable, or will a cheaper model do?" That single habit will save you more than any clever prompt trick. Each one, teach one. If this helps you get more out of Fable 5 without the limit anxiety, pay it forward. Drop your before-and-after in the comments, and if you tune it or catch something I missed, share it here so the whole room levels up!🤝

#AIS Challenge Used trigger.dev to build a Job Searching Automation, which will search jobs with specific filters based on location This automation will search jobs every Monday Morning and save it to the Supabase Database in a table. (You can see two tables named weekly_reports and seen_jobs) Jsearch Tool (Tool I used to search jobs) API key wasn't working. Claude gave me error several times, I also look for correct API. But found the same API Key every time I searched Jsearch Tool. At the end, Claude found a small version error which was causing it to break. (Only Because of Claude Code) 😉 Tools Used: Claude Code Pro, VS Code, Jsearch, Supabase( Tool for cloud database)

What’s good, everyone? I’ve been off the grid for a minute. Here's a short loom that explains my Day-5 build in a minute: https://www.loom.com/share/5274384c30e24924bbbb25741876182d I got completely fed up with manually filtering through hundreds of job leads, staring at screens, and writing the same proposals. It takes hours, and honestly, who has the time? I wanted to touch grass, so I built a fully autonomous engine to do the heavy lifting. Here is what I just cooked up: The Real-Time Lead Scoring Dashboard. I didn't just string a couple of basic prompts together; I built a legit two-tier AI architecture to avoid rate limits and keep costs at the absolute floor. - The Vibe Check: Incoming data is processed by a lightning-fast Llama 3 model running on Groq hardware. Its only job is to grade the lead on a 1-10 scale in literal milliseconds. - The Heavy Lifter: If the lead passes the vibe check, it gets routed to a massive 70B-parameter model that drafts a highly tailored, bespoke proposal. - The Aesthetics: I piped all of this backend logic through Supabase Realtime WebSockets into a custom Next.js UI styled with Tailwind and Framer Motion. The dashboard updates instantly as the AI works—no page refreshes, just smooth animations. Why this is a massive W if you run a business: I built this specific dashboard for my own pipeline, but the underlying engine is completely universal. If your company is manually qualifying inbound leads, sorting data, or dealing with clunky AI tools that crash when you try to scale, you are burning cash and time. This two-tier architecture drops API costs by 90% and runs entirely hands-free. I just recorded a Loom video breaking down the exact Next.js and n8n architecture. Drop a comment below if you want the link (or if you want me to send over the local deployment files so you can test the code on your own machine!). If you’re a founder looking to build this exact real-time infrastructure into your operations—my DMs are open. Let's chat.