Can a single Notion node in n8n retrieve data from multiple databasesat once, or do I need one node per source?

Questions
1

Hi everyone: I am trying to retrieve data from multiple tables at once with one node but seems like it only allows to retrieve one table per

Describe the problem/error/question

I am trying to have all my job experience in different tables and connecting them through referenced fields.
I know that is easier for LLMs to retrieve structured data but I am wondering if it’s possible that:
1 - The same node pull information from different tables in Notion not just one.
2 - My AI Agent use the ‘tool’ context to get different tables and understand how they are connected
In case it is not possible, I am backing to the plain text way.

What is the error message (if any)?

Please share your workflow

Share the output returned by the last node

Information on your n8n setup

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app): n8n cloud
  • Operating system: MacO Sequoia 15.6.1
2

hi @Alfredo_IM

these templates will help you

4

Hi @Alfredo_IM

You cannot use a single Notion node to pull data from multiple tables at once because each node is designed to talk to only one specific table ID. If your job experience is spread across different tables, one node simply isn’t built to “hop” between them in a single step.

However, your AI Agent can solve this by using a “toolbox” approach. Instead of one node, you give the Agent multiple tools—one for each of your Notion tables. By writing clear instructions in the system prompt, you can teach the Agent to first find a piece of information in one table and then use that result to search for connected data in another.

For the most reliable results, the best method is to create a separate n8n workflow that does all the “joining” of your tables and merges the data into one clean summary. You then connect this entire workflow to your AI Agent as a single tool. This ensures the data is perfectly organized before the AI even sees it, preventing the Agent from getting confused by the table connections.

See below for example.

Part 1: The Data Joiner (Sub-workflow)

This workflow does the “heavy lifting.” It takes a search term, finds the company, finds the related jobs, and cleans the data.

Part 2: The AI Agent (Main Workflow)

This is the interface. It uses the “Workflow Tool” to call the sub-workflow above.

Important: Because I do not have your specific Notion Database IDs or Property Names, I have used placeholders like YOUR_COMPANY_DB_ID and Company Name. You will need to click into the Notion nodes and select your actual databases from the dropdown menus.

5

hey @Alfredo_IM One node, one database, that is a hard limit. Each Notion node targets a single database ID, confirmed in the n8n notion node document : Notion node documentation | n8n Docs

You need one Notion node per database, then merge the results with a Merge node before passing to your agent.
For your AI Agent setup, add each database as a separate Notion tool, the agent will query them as needed and can reason about the relationships from the returned data.
Do not go back to plain text. Clean structured data per table is exactly what makes LLM retrieval accurate.
If your tables use Notion relation fields, retrieve the parent first, extract the relation IDs, then feed those into the next Notion node to pull linked records.

6

Short answer: no, one Notion node reads one database. The Notion API itself is per-database for queries, so n8n cannot do it in a single node, it is a platform limit, not an n8n one.

The clean pattern for your case (job-experience tables linked by referenced fields): one Notion node per database to pull each, then a Merge node to combine them on the matching reference field, then pass the joined result to your AI Agent. If you have many tables, loop: a Split or a list of database IDs into a single Notion node run once per ID, collecting results, then merge.

One thing for the LLM-retrieval goal you mentioned. Rather than have the agent query Notion live every time (slow, and more calls that can fail), consider pulling all the tables on a schedule, flattening them into one combined store, and pointing the agent at that. Fewer moving parts at query time, and if Notion has a hiccup your agent is not dead in the water. How many tables are you joining? That decides whether merge-nodes or a loop is cleaner.