Example prompt orchestrations to navigate complicated multi-step workflows

- This is a blog post by **Simon Willison**, a well-known British programmer and open-source advocate (creator of Datasette, co-creator of Django). He frequently writes about practical uses of large language models (LLMs) like ChatGPT and Claude. - The post demonstrates **prompt scaffolding** (or "orchestration"): structuring a series of prompts and tools to tackle complex tasks that a single prompt cannot handle reliably. This is a key technique for moving beyond simple Q&A with AI. - It builds on a growing ecosystem of "**agentic**" patterns where an LLM is given a step-by-step plan, external tools (e.g., a code executor or web search), and the ability to loop until a goal is met. - The examples shown — iterating on code, verifying facts, or chaining research steps — reflect a shift from treating AI as a chat partner to treating it as a **programmable reasoning engine** that can execute workflows.