AI is evolving from a single agent to multiple — it is in the ether — it is becoming.
Companies are making a default operation that your single Claude Code now launches multiple different agents (to do specific parts of the job and then report back) — sometimes (regularly 🙃) without you asking it to do that research.
What does this look like for me & for H3? What is needed for a product versus a person? Is this over-engineering or progress?
... this is the question for today and the next week during build.