Hey guys! I just sat down with @Patrick Chouinard for one of the coolest deep dives we've done in the community. Patrick has become our community's go-to AI expert and one of the most helpful resources we have. He's quietly been building something most teams pay a vendor $50k a year for. He's turning every community call we've ever recorded (two and a half years of two to three hour conversations) into a queryable "community brain." You'll soon be able to ask it anything that's been discussed and get a real answer with citations. Here's the part that broke my brain. Standard RAG completely falls apart on transcripts. The question gets asked at minute 6, the conversation drifts, and the real answer shows up at minute 41. A normal chunker has no idea those two moments belong to the same idea. Patrick solved it by adding an LLM analysis layer BEFORE chunking that restructures the transcript into self-contained units of knowledge. You can watch the full breakdown above! Here's everything we covered: ✅ Why traditional chunk-and-embed fails on non-linear data ✅ The LLM analysis layer that turns raw transcripts into RAG-ready knowledge ✅ How Patrick picks the cheapest model that's still smart enough for each step (Kimi K2.5 for restructuring, Sonnet 4.6 for the signal extraction) ✅ Why he chose LanceDB over Pinecone (and when you'd flip that decision) ✅ Running the whole thing locally with Ollama, Open Web UI, Gemma 4 4B, and gpt-oss:20B for more complex retrieval ✅ Using Claude Code to build a custom chunker instead of fighting with a library ✅ Real cost math. About 40 cents per two hour episode and under $100 to process the entire archive The wildest part is the price. Once the embeddings are built, querying is free forever because everything runs on your own machine. No SaaS, no per-token cost, no IT review. Patrick is also planning to open source the full pipeline once a few rough edges are ironed out. So if you've been wanting to build something like this for your own team, agency, or client, you'll have a working blueprint to start from.

Everybody’s building automations. Nobody’s asking which tasks should even be automated. I run a 50-person company. We sat down with every department and asked one question about every recurring task: does this need a human? Not “can AI do this?” That’s the wrong question. The right question is “does a human add any cognitive value here?” Take accounts payable. Bill comes in as a PDF email. Someone opens it on one screen, copy-pastes it into the accounting system on another screen, then emails it to the bank for payment. That’s three screens. Zero thinking. The system should be doing that. You should be upset that you’re doing it. That’s bucket one. Digital. 42% of our tasks landed there. Bucket two is judgment. Things where AI can prep the work but a human has to make the call. Vendor disputes. Ambiguous invoices. HR issues. 35% of tasks. Bucket three is the point of all this. Contributor. The stuff your people DON’T do right now but COULD do. Ideas. Process improvements. One of my team members suggested attaching plain-English FAQs to our technical quotes so customers actually understand what they’re buying. Simple. Nobody thought of it because everyone was too busy copy-pasting invoices. I published the whole framework. 90-day rollout, the psychology behind why people resist, daily playbook, real case study with numbers. https://3buckets.ai What’s the most mind-numbing task in your business that a human is still doing for no good reason?

📎 SHARED RESOURCES RecapFlow Documentation Site (Patrick's automated meeting recap project) https://recapflow.patchoutech.com/ Assay AI (Ty's hallucination-reduction project) https://tryassay.ai OpenArt Suite (multi-model image and video generation) https://openart.ai/suite/home AI Engineer YouTube Channel (agentic systems, YC-equivalent content) https://www.youtube.com/@aiDotEngineer Nate Jones YouTube: Choosing Coding Models (highly recommended) https://www.youtube.com/watch?v=09sFAO7pklo Lenny's Podcast YouTube (how companies are reorganizing in the AI era) https://www.youtube.com/@LennysPodcast Karpathy AutoResearcher Post https://x.com/karpathy/status/2031135152349524125 Fine-Tuning Modalities Video from AI Engineer channel (rated 10/10, presented by OpenAI employee) https://youtu.be/JfaLQqfXqPA?si=7-dSakV1q98LNsvf Kimi Code (potential Claude Code supplement or review alternative) https://www.kimi.com/code/en CMUX Terminal (gives Claude Code genuine CLI-level terminal and browser control) https://www.cmux.dev/ Symphony by OpenAI (multi-agent coordination using Linear as task board) https://github.com/openai/symphony Claude Facial Expression Analysis via Claude Code (Matt Berman YouTube) https://www.youtube.com/watch?v=cHgCbDWejIs Fieldy AI (wearable ambient audio capture with webhook and N8N integration) https://www.fieldy.ai/ NVIDIA Startup Cloud Credits Program (up to $250K in Google Cloud credits for qualifying startups) — Paul offered to post the application form in the community forum. Watch for that post.

Hey guys, I just published a detailed documentation on how I deployed and secured my instance of OpenClaw/Moltbot/Clawdbot, feel free to take a look at : https://opclwsec.patchoutech.com/ There is also a link to a shared NotebookLM Notebook in there so you can question the docs and look at the video and audio overviews. Let me know what you think, I kinda like this mode of publication over traditional "blog post"

Took the plunge and deployed it on my own AWS infrastructure. For those who've been curious but haven't pulled the trigger yet - it's real and it works. My setup: - Isolated VPC with private subnet (no public IP) - Access via Telegram only - Zero exposed ports - SSM for admin - Encrypted storage, locked-down permissions First conversation hit and Claude responded through Telegram. Wild feeling having an agent just... waiting for me. Security was my main hesitation. Solved it by putting everything behind NAT with no inbound routes. The agent can reach out (APIs, Telegram) but nothing can reach in. If you're on the fence - the infrastructure side is more approachable than it looks. Happy to compare notes with anyone else who's deployed.