All products

Load data into MotherDuck with AI

Inspect an authorised source, choose the right ingestion method and validate the target table. Decide in advance what should happen when the same input arrives twice.

To load data into MotherDuck with AI, give motherduck-load-data the source, target and expected unit of each row. A weekly branch-sales file might use dates in more than one format. Looking at a sample before committing to a column type can catch errors that would otherwise appear as missing weeks in a dashboard.

The next decision is how the load behaves on a repeat run. Appending the same file twice can double the results. A keyed update, replacement or staging step may be appropriate, depending on the source and the intended record history. The workflow checks counts, types and sample values after ingestion so the team can compare them with its starting material.

What it does

  • Inspect a file or connected source before loading.
  • Check dates, numeric fields and missing values.
  • Decide whether to append, replace or update.
  • Plan for repeat deliveries of the same records.
  • Validate the resulting table against the source.

Install and use

  1. Install the skill in a compatible agent. Live ingestion needs an authorised MotherDuck connection, permission to write the target and access to the selected source.
  2. Ask “Inspect our weekly sales file. Check the dates and amounts and explain how a repeat load would affect order counts before writing anything.”
  3. Compare the resulting count, a known total and several sample records with the source.

FAQ

Source

Made by MotherDuck. Original skill: motherduck-load-data, published on GitHub under the MIT licence.

View on GitHub