Process multiple requests in FIFO using OpenAI Batch API and Supabase/Postgres

by Teng Wei Herr · 121 views · created 2026-02-06 · Engineering, AI Summarization

What this workflow does

This workflow has 16 nodes and runs manually, when you click “Execute workflow” (plus 1 other trigger). Along the way it connects Supabase and Postgres. It also calls api.openai.com over HTTP.

How it runs

Trigger: Manually, when you click “Execute workflow” (plus 1 other trigger).

1. Convert to batch requests in .jsonl (Code): runs custom JavaScript code
2. Get the earliest uncompleted batch (Postgres): select
3. Call Files API (HTTP Request): calls api.openai.com (POST)
4. Call Batch API to retrieve batch object (HTTP Request): calls an HTTP API
5. Call Batch API (HTTP Request): calls api.openai.com (POST)
6. If: checks a condition and branches into true / false
7. Create a row in batch table (Supabase): uses Supabase
8. Download .jsonl result (HTTP Request): calls an HTTP API
9. Update status (Supabase): update
10. .jsonl to base (Extract from File): extracts data from a file (binary to propery)

…plus 2 more steps. See the full node list below.

Services it uses

SupabasePostgres

HTTP calls to: api.openai.com

Before you upgrade to n8n 3.0

Nodes used (9)

• HTTP Requestn8n-nodes-base.httpRequest
• Manual Triggern8n-nodes-base.manualTrigger
• Ifn8n-nodes-base.if
• Coden8n-nodes-base.code
• Extract from Filen8n-nodes-base.extractFromFile
• Schedule Triggern8n-nodes-base.scheduleTrigger
• Supabasen8n-nodes-base.supabase
• Postgresn8n-nodes-base.postgres
• No Operation, do nothingn8n-nodes-base.noOp

From the author

How it works You provide a list of prompts and a system instruction, the workflow batches them into a single OpenAI Batch API request. The batch job is tracked in a Supabase openaibatches table. A cron job polls OpenAI every 5 minutes, and once the batch…

View the original on n8n.io → Check your own copy