A day at @AnthropicAI . @lamismukta and others showed me around Claude Managed Agents, Long-Running Agents and Claude Cowork.
A year ago the platform evolution slide would have been mostly model names and context windows. Now Anthropic is shipping the harness building blocks, the scaffolding around the model that decides what it sees, remembers, and checks. Borrowing a line from @rgb_prithvi's harness design post, every component in a harness encodes an assumption about what the model can’t do. And as models improve, those assumptions need re-examining. The harness is the product. Thanks to the team for putting it on!
Throughout history, humans have built monumental structures aligned with the stars, crafted with intricate detail, and designed to stand the test of time and with AI, we can take these construction principles 100x further blending the wisdom of the past with the limitless potential of the future.
My @openclaw just told me to get off the computer and get some rest. CTO Order. Briefing at 23:30 before I go to sleep.
Thanks @steipete. New beta feels solid so far with @GoogleDeepMind ’s gemini 3.1 .
Software engineering accounts for nearly 50% of all AI agent tool calls. Healthcare, legal, finance, and a dozen other verticals are barely touched, each under 5%. That's a hundred AI unicorns waiting to be built.
https://t.co/cdJnGqsjHM