Documentation Index
Fetch the complete documentation index at: https://parabola.io/docs/llms.txt
Use this file to discover all available pages before exploring further.
Examples of categorizing data with AI
- Clean a table where the names of columns are not always consistent, such as packing lists from multiple vendors. Convert columns with names like “Product ID,” “sku,” and “Style” into “SKU”
- Take a list of clothing items with inconsistent size names and standardize those names. Convert sizes with names like “Med,” “M,” and “mdium” into “Medium”
- Correct words that are misspelled, contain errant spaces, or have undesirable casing.
How to use this step
Selecting what to evaluate
Start by choosing whether you want to clean up the names of your columns, or the values in a specific column.- Standardize column names: the AI will rename columns based on examples that you provide.
- Standardize values in a column: The AI will clean values in a specific column using examples that you provide as a guide. The cleaned values will be displayed in a new column, next to the column that the AI evaluated
Setting up examples
This step is designed to use examples to guide the AI to find values or column names to update. The better the examples, the better the AI will be at evaluating and standardizing your data. When evaluating a value, the AI will use the examples, as well as the desired new value, to make decisions. It will also infer and guess at names that you may not have given an example for. Include as many examples in each value as you would like, and separate them with commas. Each “Value” in the step settings should correspond to a single new value that you want to use as a column name or a cell value.Fine tuning
Helpful tips
- Sometimes you’ll see a response or error instead of a result. Those responses are often generated by the AI, and can help you modify the prompt to get what you need.
- Adding more examples in each value can help guide the AI
Frequently asked questions
How many examples should I provide? Start with 2-3 examples per target value. If the AI misclassifies edge cases, add more examples or use the Fine tuning field to describe how to handle them. Will the same input always produce the same output? The AI is generally consistent for inputs that match your examples closely, but rare or ambiguous values may shift between runs. Pin behavior down by adding those values as explicit examples. Can I standardize multiple columns in one step? Each step standardizes column names or the values in one column at a time. To clean several columns, chain multiple Standardize Data with AI steps together — or pair this step with Categorize with AI and Extract with AI for related cleanup tasks.Related steps
- Categorize with AI — bucket rows into categories using AI.
- Extract with AI — pull structured fields out of unstructured text.
- Clean data — strip whitespace and apply non-AI cleanup rules.
- Find and replace — handle deterministic substitutions without AI.
- Change text casing — normalize casing across a column.
What is Prowork?
How Parabola’s AI builder works — and how AI steps fit into full workflows.
Build a Flow with Prowork
Describe what you need and Prowork constructs the steps, including AI steps.