Products

Stack Lab

Chain several AI models into one pipeline, with gates where you want a say and a measurement at every step.

What it gives you

Chain models into one pipeline. Put a gate where you want a say. Measure what each step adds, instead of trusting one model's first answer.

A stack is a chain of steps:

Writer steps draft or revise.
Scorer steps score the work on rubrics, such as importance and urgency, or impact and effort. A scorer never writes.
Rules decide what happens next: stop, finish, skip a step, or pause at a gate.

A stack can run straight through, continue only when a score is good enough, or stop at a gate and wait for a person.

Every step records its own measurements. Two stacks can be compared on the same rubrics, so you can see which one is better and by how much.

Stack Lab runs the pet's stack superpowers. A saved stack can be called from the pet's Super menu like any other superpower.

Measure each step, or you cannot tell what it adds. Every stack gets a rubric and a comparison before it is adopted.
The scorer is not the writer.
Put a person's gate only where their judgment is the point.
Check that a model is available before a stack depends on it.

Live in the 0.0.1 alpha.