Ingest and search Cloudflare R2 media with Gemini, Groq Whisper, and Supabase

by Dave Sartori ✓ · 3 views · created 2026-06-20 · Document Extraction, AI RAG

What this workflow does

This AI workflow has 49 nodes and runs when an HTTP POST request hits its webhook URL (plus 5 other triggers). Along the way it connects Embeddings Google Gemini, Supabase Vector Store, Google Drive and Google Gemini and 1 more. It also calls n8n.airpg.ai over HTTP.

How it runs

Trigger: When an HTTP POST request hits its webhook URL (plus 5 other triggers).

1. Configure Input Data (Edit Fields (Set)): sets 6 fields: bucket, prefix, bucket_url…
2. Query Vector Store (Supabase Vector Store): searches the Supabase vector store, with embeddings: Embed with Gemini Model
3. Separate Image Metadata (Split Out): splits a list into separate items
4. Separate New Image Output (Split Out): splits a list into separate items
5. Separate PDF Metadata (Split Out): splits a list into separate items
6. Separate New PDF Output (Split Out): splits a list into separate items
7. List R2 Storage Files (Cloudflare R2Storage): list
8. Rank Search Results (Edit Fields (Set)): sets 4 fields: public_url, mime_type, summary…
9. Submit Image to Gemini API (HTTP Request): calls an HTTP API (POST)
10. Parse Image Output (Edit Fields (Set)): sets 10 fields: album, public_url, file_date…

…plus 29 more steps. See the full node list below.

Services it uses

Embeddings Google GeminiSupabase Vector StoreGoogle DriveGoogle GeminiCloudflare R2 Storage (community)

HTTP calls to: n8n.airpg.ai

Before you upgrade to n8n 3.0

Nodes used (17)

• Edit Fields (Set)n8n-nodes-base.set
• Embeddings Google Gemini@n8n/n8n-nodes-langchain.embeddingsGoogleGemini
• Default Data Loader@n8n/n8n-nodes-langchain.documentDefaultDataLoader
• Character Text Splitter@n8n/n8n-nodes-langchain.textSplitterCharacterTextSplitter
• Supabase Vector Store@n8n/n8n-nodes-langchain.vectorStoreSupabase
• HTTP Requestn8n-nodes-base.httpRequest
• Webhookn8n-nodes-base.webhook
• Coden8n-nodes-base.code
• Mergen8n-nodes-base.merge
• Filtern8n-nodes-base.filter
• Split Outn8n-nodes-base.splitOut
• Google Driven8n-nodes-base.googleDrive
• Respond to Webhookn8n-nodes-base.respondToWebhook
• Aggregaten8n-nodes-base.aggregate
• Google Gemini@n8n/n8n-nodes-langchain.googleGemini
• Cloudflare R2Storagen8n-nodes-cloudflare-r2-storage.cloudflareR2Storage
• Switchn8n-nodes-base.switch

From the author

Quick overview This workflow ingests images, PDFs, and videos from a Cloudflare R2 folder, uses Google Gemini to view pdfs, images and videos, Groq stt (Whisper) for video transcriptst - to generate searchable descriptions and tags, stores embeddings in a…

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