Adding RAG: Chat with Documents Locally
A beginner-friendly guide to local document chat, embeddings, vector databases, Open WebUI knowledge bases, privacy, and common RAG mistakes.
Open WebUI Deep Dive: Users, Models, Documents, Prompts, and Safe Sharing
Configure Open WebUI accounts, model and document access, provider connections, backups, and updates, with controlled signup and restricted tools before sharing an instance over LAN or VPN.
OpenWrt 24.10.8 and 25.12 Migration Runbook: Patch, Backup, Sysupgrade, and Recovery
An OpenWrt router upgrade runbook for patching exposed services, preserving config, checking device support, testing DHCP, Wi-Fi, DNS, and keeping recovery ready.
Keycloak 26.7.1 Security Response: Map the Fixes to the Features You Actually Use
Keycloak 26.7.1 includes many security fixes. This guide maps them to OIDC, SAML, FGAP, DCR, LDAP, metrics, and upgrade validation.
Local AI Hardware Sizing: CPU, NPU, GPU, RAM, and VRAM
A local AI hardware sizing guide explaining model memory, quantization, KV cache, CPU-only use, NPUs, GPUs, RAM, VRAM, and upgrade tiers.
Building the Right PC for Local AI
A plain-English guide to choosing the right CPU, RAM, GPU, VRAM, storage, power supply, and cooling for a local AI homelab PC.
Bazarr 1.6.1 Security Response: Patch or Isolate, Rotate Keys, and Check for RCE
For affected Bazarr builds, isolate untrusted access, preserve evidence and verify a current fixed release. Review exposed secrets and logs before reopening access, distinguishing reasons to investigate from evidence that warrants rebuilding.
Portainer 2.44 Security and Bootstrap Tokens: Secure First Run and Existing Installs
A Portainer security runbook for first-run exposure, setup tokens, agent trust, endpoint permissions, backups, and management-plane isolation.
Home AI Hardware Levels: A Beginner's Guide to the PC You Need
A plain-English guide to choosing hardware for a home AI setup, from using an old PC to building a serious GPU workstation.
Local AI Models Explained: Gemma, Llama, Qwen, Nemotron, Phi, DeepSeek, and More
Decode local model names, quantization, context, and memory requirements, then compare exact checkpoints on your own tasks rather than choosing by family name or advertised size.