Learn from the structure she built and apply it to your own obsidian or your own built-in second brain. I'm sharing this because I like the structure how she has it. See attached https://www.youtube.com/watch?v=LGwv4qXGgEo

Just read this article and wondering if anybody tried it out. Apple Silicon Macs running macOS 26 (Tahoe) ship with an on-device language model as part of Apple Intelligence. It's a roughly 3 billion parameter model that runs on the Neural Engine, works offline, and never sends data to a third party. The catch is that it's only accessible through Apple's FoundationModels Swift framework, which means building a native macOS app just to use it. Apfel is an open-source tool that removes that barrier, exposing the model through a CLI, an interactive chat mode, and an OpenAI-compatible local server. Read the full article here: https://betterstack.com/community/guides/ai/apfel-device-llm/

The count, from my own bench: every clip I've shot or streamed was searchable by exactly one thing. Its filename. Anything silent, anything visual, anything nobody said out loud: gone. Search what was said, lose what was shown Transcript pipelines only index speech. The take where the red jacket walks left, the quiet screen moment, the beat the chat ignored: none of it survives, because nobody said it out loud. This isn't magic, and it isn't cloud. Everything runs on my own hardware, and the first release candidate proves plumbing before it promises understanding. Query without re-seeing The fix is a four-stage spine. Sample the clip on scene cuts and motion, not every frame. Observe it once with a local vision model and write versioned sidecars. Index those sidecars into full-text search plus embeddings. Then query: ranked time spans, no vision re-run, ever. A five-second test clip ingests in 0.38 seconds. A gold harness scores recall on every change, so quality is a number, not a vibe. The release candidate is yours ODVM v0.1.0-rc1 is up for ARCHITECT members: pip install, four commands, your footage stays on your disk. Break it and tell me where. The model is replaceable. The memory you build is the asset. Local first. Query without re-seeing. The discipline beats the tool. Full deep-dive with diagrams on aris-space: https://aris-space.com/documents/tools-and-plugins/odvm-video-memory //A<3

Two Engineers on Why Your Agent Demo Will Not Survive Production

📣 I enjoy finding truth and education in humor. 🟢 See all 10 live in the classroom now: https://www.skool.com/cliefnotes/classroom/b5143824?md=367fe0cb353f44f28e1eb493dbe40054 ❓Did you learn something new? If you have a better one comment it below 👇