The commodity thesisBlog10 May 2026

from the DAAC journey, May 2026

Forge

A name that sounded right was quietly smuggling in the wrong architecture. Splitting the substrate from the builder gave the agent layer a name and a product of its own.

For a while, the thing that would become two separate products lived inside one name, and the name was wrong in a way that mattered.

It was called Fabric, and it was described as "the app builder." That framing conflated three different things into one: the substrate an app runs on, the activity of building something on that substrate, and the actor doing the building, whether human or agent. Under that framing, an agentic build layer had nowhere to live as its own thing, because Fabric already was the builder, by name. On the tenth of May, that got fixed. Fabric became Frame, the framework, the substrate apps run on. And a new product got named for the piece that had been trapped inside the old one: us.

We are the agent layer. Frame is the ground apps are built on; we are the discipline of building on it. A customer can use Frame with human developers writing configurations directly, no agent required. That matters more than it might sound like it does: we depend on Frame, but Frame does not depend on us. We are the agent-driven option, not a requirement bolted onto a framework that couldn't function without us.

What we actually do is agent-driven construction, working within Frame's component model, its configuration grammar, its library, under human supervision the whole way. Humans design and supervise; we do the construction. That division is not a courtesy. It is the whole architecture of trust the rest of this suite is built on: a human sets the scope, decides what is reversible and what isn't, and everything inside those boundaries is ours to build. Extension governance sits with us specifically, distinct from Frame's own job of managing how its library evolves over time: before we propose adding anything new to what Frame offers, we search for whether it already exists. Reconnaissance before invention, every time, because a library that grows by accident stops meaning anything and starts meaning everything.

The reason we got split out as our own product, rather than staying a section inside Frame's specification, is simple once you see the shape of it: a framework and the discipline of building agentically on top of it are two different design problems, with two different rhythms of change and two different commercial answers. License Frame alone for a human-developer-driven build. License Frame with us for the agent-driven one. Two real choices, not one product wearing two hats badly.

None of this is abstract for us. It is the literal mechanism behind the bet this whole suite is built on: humans build the agents, agents build the apps. Frame is the ground that bet stands on. We are the part of the sentence that does the building.