We’re building TMT to make working with agents feel more like working together.
It starts in the terminal: give an agent a name, send it a task, and keep its reply connected to the work. Colab adds a visual place to look at a page, point to a detail, and discuss what should change.
Our aim is simple: make intent clear, keep context close to the work, and leave important decisions with people. Agents help us write, investigate and build. People set the direction and decide what is ready to share.
This journal is where we’ll share small experiments, working examples and lessons from building TMT—what works today, and what we’re still figuring out.
Thanks for stopping by. We’re glad you’re here.