Hey guys, I'm trying to figure out if Obsidian is worth adding to my stack. Been seeing it everywhere for a while now, everyone's using it as memory for Claude, which makes sense, it's just markdown folders. Thing is, I feel like I've already got my own version: one HQ folder of markdown files Claude reads at the start of every session, memory file, context file per project. Curious what you guys are doing. Anyone running Obsidian over an already decent folder setup? Any value add I'm missing? Cheers

It reads your Claude/Cursor/Codex sessions and profiles how you actually build with AI, runs locally so your code stays on your machine. Gives you a "builder archetype" off the back of it. Reckon this is the direction things are heading, AI being able to read how someone codes and figure out what kind of dev they are. Feels like that's going to be a big part of hiring and interviews soon, which is smart and equally snoopy. https://paxel.ycombinator.com/ Anyone run it? Worth the time, or is the profile a bit of a gimmick?

I pulled together the main options from a recent tool breakdown, plus one extra cursed-but-powerful category: building the whole thing in VS Code as HTML. I'll leave the link to the original tool breakdown in the comments.

Hey all looking for architecture advice from anyone who's been here. I'm automating the back-office side of a sales role on Claude — AR/collections, order entry, statement sends, reorder follow-ups, and more workflows being added. The setup is skill-based: I invoke workflows by command (e.g. "Submit orders"), Claude handles the steps, stops for my review before anything sends or submits. Building toward full orchestration with dispatch routing the right workflow automatically based on triggers. Problem: I'm on Max 5x and a few planning chats plus ONE order through Cowork already hit ~26% of my usage before reset. That math doesn't scale once everything is live. The bigger issue: Cowork is currently slower than submitting orders manually. It runs a screenshot-and-click loop through a web UI, and the latency is real. Has anyone found a way to speed this up or get off the vision loop entirely? Questions: 1. Is moving automated execution to the Agent SDK / metered API the right architecture — keep the subscription for interactive work, run agents on the credit pool? 2. For UI-based tools with no API access, did you find a way off the screenshot loop or just accept the cost and speed hit until a native integration exists? 3. Self-capping usage meters on subagents — is that SDK-only or doable another way? 4. Should I be building workflows in Cowork first then graduating to Claude Code, or is it worth starting in Code now? Trying to get the architecture right before I scale this up. Appreciate any wisdom thanks fam.

Everyone's stacking AI experts into "councils" right now. Here's what nobody mentions: most of them produce blander advice than a single good prompt. I've been building multi-agent systems for a while, and the council pattern is seductive. Load six marketing legends, let them debate, synthesize the genius. Three things I learned the hard way: 1. Councils regress to the mean. Put Cialdini, Godin, and Vaynerchuk in a room and "synthesize" their answers and you get generic marketing advice wearing three nametags. The fix isn't a better synthesizer. Stop resolving the disagreement. Let the tension stand and make one agent own the call. 2. The debate is where your budget dies. Distilling a book into a tight skill file is cheap. Having agents argue in real time is not. If "minimal tokens" is your pitch, the preprocessing is doing the work and the roundtable is the luxury. 3. It doesn't make the model smarter. Cold Claude already does a soft version of all of this. What the structure buys you is named, sharp, predictable behavior. Say that honestly — the moment you claim it makes the AI "smarter," you've oversold it. None of this means don't build councils. It means build them with your eyes open. The real test for any council: do your experts actually disagree, or do they just agree in different vocabulary? If it's the second, you built one expert and gave it six hats. (Riffing off the systems thread from @Curtis Hays that @David Vogel highlighted for us and the 'systems' build — good work worth pressure-testing.)