PKM tools, methods, and self-hosted wikis compared.
Personal knowledge management spans Obsidian, Logseq, DokuWiki, Zettelkasten, and PARA — the right choice depends on whether you want a local note graph, a self-hosted wiki, or an outliner-driven workflow.
Claude subscriptions no longer power agents
The quiet loophole that powered a wave of agent experimentation is now closed.
Self-hosted AI search with local LLMs
Vane is one of the more pragmatic entries in the “AI search with citations” space: a self-hosted answering engine that mixes live web retrieval with local or cloud LLMs, while keeping the whole stack under your control.
Install TGI, ship fast, debug faster
Text Generation Inference (TGI) has a very specific energy. It is not the newest kid in the inference street, but it is the one that already learned how production breaks -
llama.cpp token speed on 16 GB VRAM (tables).
Here I am comparing speed of several LLMs running on GPU with 16GB of VRAM, and choosing the best one for self-hosting.
RTX 5090 in AU is scarce and overpriced
Australia has RTX 5090 stock. Barely. And if you find one, you will pay a premium that feels detached from reality.
Remote Ollama access without public ports
Ollama is at its happiest when it is treated like a local daemon: the CLI and your apps talk to a loopback HTTP API, and the rest of the network never finds out it exists.
Compose-first Ollama server with GPU and persistence.
Ollama works great on bare metal. It gets even more interesting when you treat it like a service: a stable endpoint, pinned versions, persistent storage, and a GPU that is either available or it is not.
HTTPS Ollama without breaking streaming responses.
Running Ollama behind a reverse proxy is the simplest way to get HTTPS, optional access control, and predictable streaming behaviour.
RAG embeddings - Python, Ollama, OpenAI APIs.
If you are working through retrieval-augmented generation (RAG), this section walks through text embeddings in plain terms — what they are, how they fit search and retrieval, and how to call two common local setups from Python using Ollama or an OpenAI-compatible HTTP API (as many llama.cpp-based servers expose).
Push URL updates to search engines after deploy.
Static sites and blogs change whenever you deploy. Search engines that support IndexNow can learn about those changes without waiting for the next blind crawl.
Serve open models fast with SGLang.
SGLang is a high-performance serving framework for large language models and multimodal models, built to deliver low-latency and high-throughput inference across everything from a single GPU to distributed clusters.
Install Kafka 4.2 and stream events in minutes.
Apache Kafka 4.2.0 is the current supported release line, and it’s the best baseline for a modern Quickstart because Kafka 4.x is fully ZooKeeper-free and built around KRaft by default.
Hot-swap local LLMs without changing clients.
Soon you are juggling vLLM, llama.cpp, and more—each stack on its own port. Everything downstream still wants one /v1 base URL; otherwise you keep shuffling ports, profiles, and one-off scripts. llama-swap is the /v1 proxy before those stacks.
Most local AI setups start with a model and a runtime.
What actually happens when you run Ultrawork.
Oh My Opencode promises a “virtual AI dev team” — Sisyphus orchestrating specialists, tasks running in parallel, and the magic ultrawork keyword activating all of it.
