Primary sources
Foundational papers, standards, books, and frameworks defining the Agent Economy.
OpenAI
Defines agentic AI, catalogues emerging risks, and proposes concrete practices for developers, deployers, and users of increasingly autonomous systems.
Stanford HAI
The definitive annual benchmark on AI capability, investment, policy, and public perception — including a dedicated chapter on agentic systems.
Anthropic
The open standard for connecting AI assistants to the systems where data lives — the emerging communication layer of the Agent Economy.
Yao et al., Princeton & Google Research
The foundational prompting technique combining chain-of-thought reasoning with tool use — a cornerstone of modern agent design.
Meta AI
Self-supervised approach that teaches LLMs which APIs to call, when, and how — a precursor to agentic tool orchestration.
NIST
Voluntary framework guiding the mapping, measurement, management, and governance of risks introduced by AI systems.
Stuart Russell
UC Berkeley professor's landmark case for rebuilding AI on the principle of provable deference to human preferences.
Google DeepMind
A foundation world model that generates interactive, controllable virtual environments from a single image or text prompt.