Essay2026-09-26
From Model Monopoly to an Ecology of Expertise
Why broad frontier capability is not the same as dependable expertise — and why AI needs a division of labor built around accountable specialist systems.
Read pieceEssays & Posts
Long-form writing from Nacho Labs on agent systems, workflow design, and the operational edges that actually matter.
Essay2026-09-17
An Assistant Should Be Easy to Leave
Managed assistants can be useful. The non-negotiable part is keeping your data, history, and exit path under your control.
Read piecePost2026-08-05
Server-Hosted Coding Workspace
Why moving the coding workspace off local laptops and onto a server changed the day-to-day value of coding agents more than a new model ever did.
Read pieceEssay2026-07-06
The Agent Control Plane Is the Real Product
Why serious agent software wins or loses on routing, policy, handoffs, and operator control — not on a pretty prompt wrapper.
Read pieceEssay2026-06-23
Context Is Not Memory
Why agent teams keep using memory as a catch-all term, and why that naming mistake turns into an architecture mistake.
Read piecePost2026-06-09
Most AI Workflows Fail at the Human Boundary
The real break in many AI systems is not the model. It is the handoff between useful output and accountable human judgment.
Read pieceSubscribe
Get new essays without stalking the site
Join the Nacho Labs email list for new essays and occasional updates. Low volume. No drip-funnel nonsense.
Opens the Mailchimp signup confirmation in a new tab. Prefer not to use email? The RSS feed stays available.