Algorithm Discovery

Finding datasets, recipes, and operations by example — search by shape, data profile, or usage pattern

Data Shape Search

Algorithm Discovery lets you search across all your datasets and recipes by data profile — not just name or metadata. Search by column count, data types, value distribution, or schema similarity. Find datasets that match a particular shape, or recipes that produce a specific kind of output.

Usage-Based Discovery

Recipes are ranked by usage frequency and recency. Frequently used recipes appear higher in discovery results. The system tracks which datasets feed which recipes, building a dependency graph that lets you trace data lineage forward and backward — from source file to final chart.

Column Profile Matching

When you select a dataset, the Workbench analyses its column profiles — value ranges, cardinality, null percentage, type, and distribution shape — and suggests recipes that have been applied to similar datasets. This helps you discover workflows that worked for others (or for your past self) on similar data.

Discovery, not search. Algorithm Discovery complements the standard search box. Use search when you know the dataset or recipe name. Use Discovery when you're exploring — "show me datasets that look like this one" or "what recipes have others used on tables with a date column and a numeric measure?"
← Recipe Export 📋 Contents AI-Guided Analysis →