All products

Build an n8n data table with AI

Design simple persistent storage for a workflow, choose an external key and test whether a repeated input updates the intended row.

To build an n8n data table with AI, decide what must survive after the current run ends. Keeping a shop’s order identifier can help a later run recognise an order it has already received. The n8n-data-tables-official guide covers column types, lookups, updates and upserts while keeping external identifiers distinct from system-managed columns.

A separate table is useful when you need to inspect or update state across runs. Simply removing duplicates from a list may not require persistent storage. Test with invented records first and submit the same external key twice. The outcome should show whether the workflow found the existing row or created an unintended second copy.

What it does

  • Design columns for persistent workflow state
  • Keep an external order or request identifier
  • Look up a row using a stable key
  • Update an existing row when an event repeats
  • Build a small reference table for repeated values

Install and use

  1. Install the standalone skill and use an authorised n8n instance with Data Tables and suitable permissions for live changes. Supply sample field definitions and invented test records.
  2. Ask “This order event can arrive twice. Design a table for order ID and status and show how the second arrival updates the same row.”
  3. Test the lookup and update in a test table and inspect the selection before applying bulk changes.

FAQ

Source

Made by n8n. Original skill: n8n-data-tables-official, published on GitHub under the Apache-2.0 licence.

View on GitHub