Skip to content

Home/Blog/Building with AI

Building with AI

18 articles

July 19, 2026 Beyond vector search: how RAG actually works in 2026 RAG stopped being "vector database plus a language model" a while ago. Here's how retrieval-augmented generation actually works now, chunking, embeddings, reranking, knowledge graphs, and… July 15, 2026 Your RAG system isn't hallucinating. It never found the answer. When a RAG system gives a bad answer, almost everyone blames the model. Usually the right passage was never retrieved, and… July 12, 2026 How AI models are trained: from raw text to a system that helps A language model isn't programmed, it's grown, in stages, from a firehose of text into a system that answers helpfully.… July 10, 2026 RAG vs fine-tuning: the decision, honestly The most common question in applied AI, answered without the hedging: when do you use retrieval, when do you fine-tune,… July 9, 2026 What is fine-tuning? How to specialize an AI model Fine-tuning takes a general-purpose model and adapts it to your specific task by continuing its training on your examples.… July 5, 2026 When not to use an agent GPT-3.5 inside a structured workflow scored 95.1% on a coding benchmark. GPT-4 running free scored 67%. The structure was… July 3, 2026 How LLM inference works: why it's bound by memory, not compute Buying a faster GPU often does not make an LLM generate text any faster, and the reason is one of the more… July 2, 2026 Do you actually need a vector database? Vector databases became the default first purchase for anyone building with AI. For most projects they're the wrong first… June 29, 2026 How quantization shrinks AI models without breaking them A 70-billion-parameter model needs about 140 GB of memory at full precision. Your laptop has 16. Quantization is how the… June 27, 2026 What an agent costs, and where the money actually goes Token prices fell 67% in a year and enterprise AI bills rose anyway. The reason is a multiplication nobody models: a… June 27, 2026 Speculative decoding: faster LLM generation, same output There is a way to make a large language model generate text two to four times faster while producing output that is… June 16, 2026 Knowledge distillation: how small models learn from big ones The small, fast AI models you run on a laptop or serve cheaply to millions of users were often not just shrunk from big… June 7, 2026 What is synthetic data? Training AI on AI-made data Faced with running out of human text to train on, AI labs increasingly train models on data the models generate… June 4, 2026 What is a mixture of experts (MoE)? Bigger, cheaper AI Frontier AI models now hold hundreds of billions or even trillions of parameters, yet stay affordable enough to run at… May 30, 2026 How much energy does AI use? Training vs inference The energy cost of AI is discussed constantly and measured badly. The per-query figures in circulation are stale and vary… May 29, 2026 Why does AI need GPUs? Parallelism and the memory wall AI runs on graphics chips because of a historical accident: neural networks turned out to need the same kind of arithmetic… May 27, 2026 Run an LLM locally: how much VRAM do you need? Whether a model runs on your machine is not a mystery. It is one line of arithmetic: parameters times bytes per parameter,… May 26, 2026 Open weights vs open source AI: what's actually released Almost every model marketed as open source AI is not open source. You get the finished weights under a licence, not the…

Other subjects

Foundations 27 Theory, history and the parts that will still be true in twenty years. Language & meaning 21 How language models handle language, and the places they do not. Agents 11 Building agents, deploying them, and why most attempts do not reach production. Evaluation & evidence 12 Measurement, benchmarks, and how to tell whether a claim is worth anything. Safety & governance 11 Alignment, security, regulation and who is accountable when it goes wrong.