Lab

A clearer data pipeline makes the decision layer easier to trust.

This experimental page is now static-first and explanatory instead of depending on heavy motion or a hidden interaction model.

Lab

Pipeline stages

Input, normalize, route, persist.

The pipeline should be readable.
The output should be inspectable.
The fallback should remain obvious.

Stages

The pipeline becomes easier to reason about when the stages are visible.

Stage 1

Input

Capture raw context, metadata, and the minimum information needed to reason about the task.

Stage 2

Normalize

Structure the data so the next step can inspect it without guessing at the shape.

Stage 3

Route

Send the record to the right path based on the decision logic.

Stage 4

Persist

Keep enough state to support review, handoff, and follow-up.

Next step

Keep the experiment useful, but keep the funnel clearer.

The lab still helps explain the thinking, but it no longer competes with the primary pages.