LLMs Are Hitting a Wall. Here's Why I'm Not Worried
My take on diminishing returns in LLM scaling, why goal-mode looping might get us to half-AGI sooner than expected, and why I'm betting on mechatronics as the next frontier.
My take on diminishing returns in LLM scaling, why goal-mode looping might get us to half-AGI sooner than expected, and why I'm betting on mechatronics as the next frontier.
How I set up a self-hosted AI agent with Hermes. Custom MCPs bridging it to my personal Life OS tools, Home Assistant integration, persistent memory, and Telegram chat. Full control, no subscriptions.
Bought a broken MX Master 3s on Vinted, replaced the faulty left click button with a soldering iron and an AliExpress part. Total cost: 7,000 Ft instead of 40,000 Ft.
Build a multi-port USB-C charging station from an old PC PSU and a 3D printed case: 8x USB-A QC 3.0 ports, 1x 65W USB-C PD, and a lab bench power supply for under $40.
A step-by-step journey from a single Raspberry Pi to a multi-node homelab running Proxmox, Docker, and dozens of self-hosted services, including Nextcloud, Home Assistant, and more.