A Reading List for Claude Builders
Building well with Claude is not only an engineering problem. It is a question of how language models actually work, why alignment matters, where the field came from, and where it is heading. This page collects the books that have made us better at all of the above. Some are technical, some are policy, some are biography and history. They share one quality: they reward rereading.
This list is curated, not exhaustive. We add a book only after at least one editor has read it cover to cover and is willing to defend it in print. We do not run affiliate links here. If we recommend a book, it is because it is worth your reading time, not your click.
Shelves We Are Building
📖
Foundations of LLMs
The textbooks and long-form explainers that demystify transformer models, training pipelines, and the mechanics of in-context learning. Useful whether you write CUDA kernels or just want to know what is happening under the hood.
🧭
Alignment & Safety
From early agent-foundations work to current research on RLHF, Constitutional AI, interpretability, and scalable oversight. Reading these makes Anthropic's design choices in Claude legible.
🏛️
Policy & Society
How governments, regulators, and civil society are thinking about frontier AI. EU AI Act, US executive orders, voluntary commitments, and the disagreements that shape what gets built and shipped.
📜
History of AI
From symbolic AI through the neural-network renaissance to today's frontier labs. Understanding the prior winters and breakthroughs sharpens your intuition about what this moment actually is, and is not.
🛠️
The Craft of Building
Software engineering with LLMs in the loop. Books on agent design, evaluation, distributed systems, and the operational disciplines that production AI products demand.
👤
People & Companies
Anthropic's founding story, OpenAI's history, DeepMind's research arc, and the lives of the researchers whose ideas are baked into the models we use every day.
Coming Soon
We are reviewing the first batch of titles and will publish the initial list within the next few weeks. Each entry will include a short editorial pitch, the level of background it assumes, and an honest verdict on whether you should read it now or wait. If you want to recommend a book for consideration, get in touch via the channels on our About page.
Until the shelves fill up, the News overview and the Models page are the freshest reading we can offer.
