Tidepool’s combination of LLM and embedding technology allows you to ask questions of massive text datasets in a way that wasn’t possible before.
Create an attribute with a short text description to represent a “question” you would like to ask of the data.
Visualize clusters in the dataset. Inspect individual snippets of text in each cluster and create categories of an attribute that you’d like to track in the dataset.
Tidepool trains a lightweight embedding classifier for your attribute that categorizes text examples in your entire dataset and new data going forward.
Explore the distribution of your attribute and categories across the entire dataset. Chart the correlation of your attribute against metadata like business metrics. Dig into individual examples within each category.
Once you’re done exploring your data in Tidepool, export selected rows into a CSV for ad-hoc analysis or write back a table to your data warehouse so you can analyze it with the rest of your BI stack.
