Thirty-one sources in five rooms, each with two lines of our own on what it says and what it changes.

I
How to think about agents and harnesses
What the frontier labs say their own models and agents can do, read for what it changes in a working practice.
INSIDE System cards and launch notes from Anthropic and OpenAI, Anthropic's guide to building agents, and the Claude prompting and evaluation documentation.
Open the shelf
II
The papers behind the instruments
The research the site's instruments rest on, from the 2017 transformer paper to the 2026 work on long-running agents.
INSIDE Four 2026 papers on long-horizon agents and harness design, two of them from Anthropic's engineering teams, and six foundational studies: attention, model judges and their bias, long context, reasoning with tools, and constitutional training.
Open the shelf
III
Courses and training from the labs and the clouds
Where the person on your team who runs the system goes to learn it.
INSIDE Anthropic Academy, OpenAI Academy, Google Cloud Skills Boost, Microsoft Learn, AWS Skill Builder, the NVIDIA Deep Learning Institute and DeepLearning.AI.
Open the shelf
IV
Standards and guardrails
The rules and frameworks an AI deployment answers to.
INSIDE The Federal Trade Commission's proposed policy statement and consumer-protection guidance, NIST's AI Risk Management Framework and its agent-security summary, and HUD's Fair Housing Act resources.
Open the shelf
V
Reading the market
The public data behind every figure in The Record.
INSIDE Freddie Mac's weekly mortgage survey, Zillow Research data and the St. Louis Fed's FRED.
Open the shelfFormerly listed
- An entry whose publisher link fails is removed here, not left to rot, with its retirement date. None retired this cycle.
Educational, not professional advice. Nothing on this page is affiliated with or endorsed by its publisher.


