Run real AI on real hardware. No cloud fees. No data leaving your machine. No rate limits. We run a 7-VM AI infrastructure stack at home — local LLMs, voice synthesis, image generation, agents, and automation. This is where we teach you how to do the same.
Openai Codex Helps Uncover & Fix A Nearly 20 Year Old Radeon Linux Driver Bug
OpenAI Codex Helps Uncover & Fix A Nearly 20 Year Old Radeon Linux Driver Bug Welcome back to Self-Hosted Tech! If you’re anything like me, your homelab is a graveyard of repurposed hardware. That old desktop PC from college? It’s now a Proxmox node. The dusty graphics card you pulled out of a closet? It’s plugged into a headless server just to give you a display output when the network goes down. ...
Ati/Amd R300 To R500 Gpus See New 2026 Opengl Linux Driver Optimizations
ATI/AMD R300 to R500 GPUs See New 2026 OpenGL Linux Driver Optimizations Welcome back to Self-Hosted Tech! If you’ve been hanging around this blog or our YouTube channel for a while, you know we have a massive soft spot for old hardware. In the homelab community, there is a unique, undeniable satisfaction that comes from taking hardware that is decades past its prime and giving it a new lease on life. We love keeping e-waste out of landfills, and we love pushing silicon to its absolute limits. ...
Darling Now Embraces Ai For Running Macos Binaries On Linux
Darling Now Embraces AI For Running macOS Binaries On Linux Hey folks, welcome back to Self-Hosted Tech! If you’ve been running a homelab for any amount of time, you probably have a drawer (or a closet, or an entire server rack) filled with repurposed hardware. And if you’re anything like me, you’ve probably stared at an old Mac Mini or a MacBook Pro and thought, “I really want to run Linux on this, but I still need that one macOS app.” ...
Kmorph Aims To Provide Rebootless Linux Kernel Crash Recovery
Kmorph Aims To Provide Rebootless Linux Kernel Crash Recovery If you have spent any serious amount of time building a homelab, running bare-metal servers, or self-hosting your own local AI models, you have probably experienced the dreaded moment. You’re running a massive workload—maybe you’ve just loaded a 70B parameter LLM into your system’s RAM and VRAM, or you’re running a critical database for your smart home—and suddenly, the system goes dark. ...
Mcp Explained
MCP Explained: The Protocol That Lets Your Local AI Actually Do Things If you’ve been running local AI models in your homelab — whether that’s Ollama, llama.cpp, vLLM, or something else — you’ve probably hit the same wall everyone hits. Your model is smart. It can reason, summarize, generate, and chat. But it can’t do anything. It can’t read a file on your NAS. It can’t query your Home Assistant instance. It can’t search your notes in Obsidian or look up a movie in your Plex library. ...
Ask Hn: What'S Your Current Ai Agentic Setup
Ask HN: What’s Your Current AI Agentic Setup? (Self-Hosted Edition) If you spend any time browsing Hacker News, you’ve probably seen a variation of the thread title: “Ask HN: What’s your current AI agentic setup?” It’s a question that pops up every few weeks, and every time it does, the comments section turns into a goldmine of configurations, workflows, and creative ways people are pushing local AI to its limits. ...
Postman Silently Auto-Installs Ai Plugins And Attempts To Share All Api Data
Postman Silently Auto-Installs AI Plugins And Attempts To Share All API Data Let’s be honest: for years, Postman has been the undisputed king of API testing. Whether you were building a web app, integrating a third-party service, or poking around the APIs of your smart home devices, Postman was the tool you opened. It was reliable, feature-rich, and easy to use. But recently, something changed. Something that has the developer and homelab communities up in arms. Postman has started silently auto-installing AI plugins and, even worse, attempting to share your API data with their cloud-based AI models without explicit, informed consent. ...
How We Halved Dns Build Delay And Cut Memory Use By 85%
How We Halved DNS Build Delay And Cut Memory Use By 85% Welcome back to Self-Hosted Tech! If you’re running a homelab, you already know that local DNS is the beating heart of your network. It’s the traffic cop that routes your requests, resolves your internal services, and—perhaps most importantly—blocks ads and trackers at the network level. But as your homelab grows, so do your blocklists. Recently, we ran into a massive bottleneck with our local DNS setup. We were running a custom DNS sinkhole using dnsmasq inside a Docker container on a trusty Raspberry Pi 4. To keep our network squeaky clean, we had aggregated several massive blocklists—StevenBlack’s hosts, OISD, and a few custom regex lists. We were feeding nearly 6 million domains into our DNS server. ...
Ai Use In The Linux Kernel, Microsoft Defender On Linux & Other October Highlights
Ai Use In The Linux Kernel, Microsoft Defender On Linux & Other October Highlights Welcome back to Self-Hosted Tech! If you’re anything like me, you probably spent a good chunk of October swapping out hard drives, tweaking Docker compose files, and keeping an eye on the ever-evolving landscape of open-source software. October was an incredibly busy month for the homelab, self-hosting, and local AI communities. We saw some massive paradigm shifts, a few unexpected cross-platform releases, and hardware news that makes us want to re-evaluate our rack space. ...
Sleep Framework: A New Way For Llms To Retain Knowledge In Continual Learning
Sleep Framework: A New Way For LLMs To Retain Knowledge In Continual Learning Welcome back to the blog! If you’ve been hanging around the local AI scene for a while, you already know that running Large Language Models (LLMs) on your own homelab hardware is incredibly rewarding. But let’s be real: it can also be incredibly frustrating when you try to teach your model something new. ...