About

An independent brief on agentic AI.

BreakingAgent tracks what is changing in agentic AI, what matters, and what builders should do next.

What we cover

Agents are the working unit of useful AI. They reason, call tools, hold state across long horizons, and increasingly take real actions in the world on behalf of real people. The rate of change is high; the signal-to-noise ratio is low.

BreakingAgent exists to fix the second problem. We publish daily news, weekly research summaries, an actively maintained tools index, hands-on build guides, and field reports on production use cases. Reference pages cover companies, models, benchmarks, and a working glossary.

Editorial principles

  • Independent. No paid placement. No sponsored coverage. No undisclosed conflicts. Ever.
  • Signal first. Every post answers three questions: what changed, why it matters, what builders should do next.
  • Plain language. No jargon when a normal word works. We define the terms we cannot avoid.
  • Show your work. Sources cited. Methods stated. Failures admitted. Corrections in public on the corrections page.
  • Builder bias. If a story does not change what someone should ship, plan, or stop doing, it probably does not belong here.

Who we are

A small editorial team based across San Francisco, London, and Berlin, drawn from the engineering, research, and journalism sides of AI. We do not represent any vendor, lab, or investment fund.

Reach us