Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.affzero.com/llms.txt

Use this file to discover all available pages before exploring further.

What is the plan?

After you click Generate Plan, the AI produces a step-by-step transformation plan. Each step is one operation applied to your data in sequence. AffZero displays the plan in plain English before you run it. You can review it to check it matches your intent.

Plan step types

StepWhat it does
FilterKeeps only rows that match a condition (e.g. column equals a value, is in a list, contains a string)
JoinMerges two sources on a matching column
Group byAggregates rows grouped by one or more columns (sum, count, avg, min, max)
SelectKeeps only specific columns, optionally renaming them
SortOrders rows by a column (ascending or descending)
RenameRenames a single column
Steps run in order from top to bottom.

Filter operators

OperatorMeaning
=Exactly equals (one value)
Does not equal
> / < / / Numeric comparison
is one ofMatches any value in a list — use this for OR logic
is not one ofDoes not match any value in the list
containsCase-insensitive substring match
does not containDoes not contain the substring
To filter where a column matches any of several values, the plan should use a single is one of step with the full list. If you see multiple separate filter steps on the same column with =, those would be combined with AND (producing no rows). In that case, re-write your instruction to say “where X is one of [a, b, c]” and regenerate the plan.

If the plan looks wrong

The plan is shown before execution — you don’t have to run it as-is. If a step is incorrect:
  1. Edit your Instruction in section 2 to be more specific
  2. Click Generate Plan again to produce a new plan
  3. Review the updated plan
Common fixes:
  • If a filter is using the wrong column, name the column explicitly in your instruction
  • If two values that should be OR are split into two AND steps, rewrite as “where column is one of [value1, value2]”
  • If a join is using the wrong key, name the key column explicitly: “join on the ‘Affiliate ID’ column”