Skip to content
Dhanasvi
Sign in
Build an All-Source Knowledge Assistant with Claude, RAG, Perplexity, and Drive logo

Build an All-Source Knowledge Assistant with Claude, RAG, Perplexity, and Drive

Verified

AI agent retrieves and answers from Drive documents using Claude and RAG.

n8nAI & LLMAdvanced👁 176 views
Open template
Updated 2026-06-16

What this workflow does

This workflow creates a knowledge assistant that loads documents from Google Drive, embeds them for retrieval, and answers queries with Claude while preserving chat history.

It suits teams or individuals who need reliable, context-aware answers drawn from internal files and structured tool use inside n8n.

Who is this for?

Knowledge workers, researchers, and small teams who need unified answers from company files and external sources without manual searching.

What problem it solves

Information is scattered across Google Drive, chat history, and the web, making it slow and error-prone to get accurate, context-aware answers.

What it automates

Research report drafting

Pull latest data from Drive docs plus web sources to generate cited summaries in one chat.

Onboarding Q&A

Answer new hire questions using internal policies stored in Drive and conversation memory.

Competitive analysis

Combine Perplexity results with company files to compare features without switching tabs.

How the workflow works

The 13 nodes in this automation, in order.

  1. 1Google DrivegoogleDrive
  2. 2AI Agent@n8n/n8n-nodes-langchain.agent
  3. 3Embeddings OpenAI@n8n/n8n-nodes-langchain.embeddingsOpenAi
  4. 4Anthropic Chat Model@n8n/n8n-nodes-langchain.lmChatAnthropic
  5. 5Recursive Character Text Splitter@n8n/n8n-nodes-langchain.textSplitterRecursiveCharacterTextSplitter
  6. 6Call n8n Workflow Tool@n8n/n8n-nodes-langchain.toolWorkflow
  7. 7Supabase Vector Store@n8n/n8n-nodes-langchain.vectorStoreSupabase
  8. 8Default Data Loader@n8n/n8n-nodes-langchain.documentDefaultDataLoader
  9. 9OpenAI@n8n/n8n-nodes-langchain.openAi
  10. 10Postgres Chat Memory@n8n/n8n-nodes-langchain.memoryPostgresChat
  11. 11Think Tool@n8n/n8n-nodes-langchain.toolThink
  12. 12MCP Client Tool@n8n/n8n-nodes-langchain.mcpClientTool
  13. 13Reranker Cohere@n8n/n8n-nodes-langchain.rerankerCohere

Apps & integrations used

Google DriveAI AgentEmbeddings OpenAIAnthropic Chat ModelRecursive Character Text SplitterCall n8n Workflow ToolSupabase Vector StoreDefault Data LoaderOpenAIPostgres Chat MemoryThink ToolMCP Client Tool

How to set up Build an All-Source Knowledge Assistant with Claude, RAG, Perplexity, and Drive

  1. 1Import the workflow JSON into n8n
  2. 2Connect Google Drive credentials and select target folders
  3. 3Add Supabase credentials and create the vector store index
  4. 4Configure Anthropic API key for Claude Sonnet model
  5. 5Set Postgres connection for chat memory
  6. 6Activate the chat trigger and test with a sample message

How to customize this workflow

  • →Swap Anthropic model for a different Claude version
  • →Replace chat trigger with Slack or email webhook
  • →Add extra tools via MCP Client Tool node
  • →Change vector store from Supabase to another supported provider

Build an All-Source Knowledge Assistant with Claude, RAG, Perplexity, and Drive: pros & cons

Pros

  • +Combines RAG, web search, and memory in one agent
  • +Uses strong Claude model for high-quality answers
  • +Persistent memory across sessions via Postgres
  • +Direct Drive access for private documents

Cons

  • –Requires multiple paid API keys (Anthropic, OpenAI, Supabase)
  • –Setup involves several external services
  • –Advanced workflow may need debugging for tool calls
Did you find this helpful?

Frequently asked questions

It creates a chat assistant that answers questions using your Google Drive files, conversation history, and web search via Claude and RAG.

User reviews

Verified reviews from the community shape this listing's rating.

Loading reviews…

Sign in to review

Promote Build an All-Source Knowledge Assistant with Claude, RAG, Perplexity, and Drive

Add this badge to your website, or share the tool.

DFeatured on DhanasviBuild an All-Source Knowledge Assistant with Claude, RAG, Perplexity, and Drive 0