Articles
What Is a Vector Database?
Specialised infrastructure for approximate nearest-neighbour search, trading exactness for speed. Under about a hundred thousand vectors, brute force or pgvector will serve you better than a new service to operate.
What Are Embeddings?
An embedding turns text into coordinates where distance approximates similarity of meaning. Excellent at paraphrase, weak at negation and exact identifiers — and the benchmark built to rank them found no model dominates.
What Is a Context Window?
The context window is working memory measured in tokens, shared by your prompt, history, retrieved documents and tool schemas alike. Research shows models use it unevenly — strong at the edges, weak in the middle.
Why AI Models Hallucinate
Hallucination isn't a malfunction: benchmarks reward confident guessing and penalise "I don't know", so training selects for it. Purpose-built retrieval tools sold as hallucination-free still fabricate 17–33% of the time.
What Is an AI Agent?
The textbook taxonomy of five agent types won't help you ship anything. The distinction that does: in a workflow you wrote the path, in an agent the model chooses it — and everything hard about agents follows from that.
RAG vs Fine-Tuning vs Prompt Engineering
They fix different problems: prompting changes what you ask, RAG changes what the model knows, fine-tuning changes how it behaves. Research shows RAG consistently beats fine-tuning for knowledge — start cheap and escalate on evidence.
Prompt Injection: Why It Isn't a Bug You Can Patch
There is no prepared statement for natural language. Prompt injection is a property of how models work, not a defect — so the only durable defence is architectural. The lethal trifecta, what fails, and what to do instead.
How to Build an MCP Server
Most MCP tutorials predate the 2026-07-28 revision, which replaced the init handshake with server/discover and deprecated sampling and logging. What to build now, why the tool description is the hard part, and the security you can't skip.
How to Prompt Open-Source and Challenger Models
A practical guide to prompting Llama, DeepSeek, Mistral, Grok, and Qwen — chat templates, system prompt quirks, and self-hosting.
How to Prompt Google's Gemini Models
A practical, accurate guide to prompting Gemini 3 and 2.5: system instructions, long context, multimodal input, JSON mode, and thinking levels.