Large language models, prompting, Claude API, and AI-powered development
6 posts

How to give Claude tools, handle tool calls in a loop, and build reliable agentic workflows.

Semantic search over your own content using vector embeddings, pgvector, and a similarity query layer.

A practical comparison building the same content pipeline with both APIs — cost, quality, and reliability.

Real-time AI output using Vercel AI SDK + Claude streaming API in a Next.js App Router project.

Techniques to make Claude and other LLMs return reliable, typed JSON every time — without hallucinated fields.

How I used Claude to auto-generate structured blog content at scale across multiple tenant sites.