Resources
Notes, guides, and background reading
Short pieces on what a predictive layer is, how tabular foundation models work, and how to get useful results out of a small set of labels. We are writing these now.
What is a predictive layer?
Enrichment tools tell you facts that already exist somewhere. A predictive layer tells you something nobody has written down yet: your own judgement, applied to rows you have not read.
How tabular foundation models work
Why a model pre-trained on millions of synthetic tables can predict your column without a training run, and what that changes about the workflow.
Labeling well with fifty rows
How to pick examples, keep a criterion consistent, and cover the edges. The part of the job that actually decides the quality of the output.
Reading the confidence column
What the probability on each row means, where to set a review threshold, and how to spot a prediction you should not trust.
Want one of these sooner?
Tell us which and we will send it when it is ready, or just answer the question directly.