A clearer view
of AI agents.
A focused place to follow what agents can do, how they're built, and where their limits remain.
What belongs here
Agents that take action: browser and computer use, coding, research, and workplace workflows. We also cover tools, permissions and evaluation methods. General AI announcements qualify only when they materially change an agent's capabilities.
Our editorial approach
We prepare short, original briefs with direct links to the original material. Vendor claims stay attributed. We distinguish announcements, research and guidance from our own assessment.
Source dates, availability and limitations belong in each real brief. Reading an announcement is not independent product testing. We do not copy full articles or assume permission to reuse a publisher's images.
The first edition
Our first edition contains three original briefs prepared with AI assistance and reviewed before publication. Article dates and source dates are listed separately.
Automatic fetching, AI summarisation through an API and scheduled publication are not connected. The source list below is our starting registry. We will review the first seven editions before automatic publication.
How we choose stories
Check original announcements, filter for agent capabilities and remove repeated coverage of the same event. Skip stories without enough evidence. The planned daily limit is two briefs or one digest with up to three stories.
Corrections and ownership
A publisher contact address and corrections process still need to be added. No contact address has been invented.