AI-powered WhatsApp chatbot for text, voice, images, and PDF with RAG
Who is this for?
This template is designed for internal support teams, product specialists, and knowledge managers in technology companies who want to automate ingestion of product documentation and enable AI-driven, retrieval-augmented question answering via WhatsApp.
What problem is this workflow solving?
Support agents often spend too much time manually searching through lengthy documentation, leading to inconsistent or delayed answers. This solution automates importing, chunking, and indexing product manuals, then uses retrieval-augmented generation (RAG) to answer user queries accurately and quickly with AI via WhatsApp messaging.
What these workflows do
Workflow 1: Document Ingestion & Indexing
- Manually triggered to import product documentation from Google Docs.
- Automatically splits large documents into chunks for efficient searching.
- Generates vector embeddings for each chunk using OpenAI embeddings.
- Inserts the embedded chunks and metadata into a MongoDB Atlas vector store, enabling fast semantic search.
Workflow 2: AI-Powered Query & Response via WhatsApp
- Listens for incoming WhatsApp user messages, supporting various types:
- Text messages: Plain text queries from users.
- Audio messages: Voice notes transcribed into text for processing.
- Image messages: Photos or screenshots analyzed to provide contextual answers.
- Document messages: PDFs, spreadsheets, or other files parsed for relevant content.
- Converts incoming queries to vector embeddings and performs similarity search on the MongoDB vector store.
- Uses OpenAI’s GPT-4o-mini model with retrieval-augmented generation to produce concise, context-aware answers.
- Maintains conversation context across multiple turns using a memory buffer node.
- Routes different message types to appropriate processing nodes to maximize answer quality.
Setup
Setting up vector embeddings
- Authenticate Google Docs and connect your Google Docs URL containing the product documentation you want to index.
- Authenticate MongoDB Atlas and connect the collection where you want to store the vector embeddings. Create a search index on this collection to support vector similarity queries.
- Ensure the index name matches the one configured in n8n (data_index).
- See the example MongoDB search index template below for reference.
Setting up chat
- Authenticate the WhatsApp node with your Meta account credentials to enable message receiving and sending.
- Connect the MongoDB collection containing embedded product documentation to the MongoDB Vector Search node used for similarity queries.
- Set up the system prompt in the Knowledge Base Agent node to reflect your company’s tone, answering style, and any business rules, ensuring it references the connected MongoDB collection for context retrieval.
Make sure
Both MongoDB nodes (in ingestion and chat workflows) are connected to the same collection with:
An embedding field storing vector data,
Relevant metadata fields (e.g., document ID, source), and
The same vector index name configured (e.g., data_index).
Search Index Example:
{ "mappings": { "dynamic": false, "fields": { "_id": { "type": "string" }, "text": { "type": "string" }, "embedding": { "type": "knnVector", "dimensions": 1536, "similarity": "cosine" }, "source": { "type": "string" }, "doc_id": { "type": "string" } } } }
AI-Powered WhatsApp Chatbot for Text, Voice, Images, and PDF with RAG
This n8n workflow creates a sophisticated AI chatbot that integrates with WhatsApp Business Cloud to process various message types (text, voice, images, and PDF documents) and leverages Retrieval Augmented Generation (RAG) for enhanced, context-aware responses.
Description
This workflow automates an intelligent WhatsApp chatbot capable of handling diverse user inputs. It intelligently routes messages based on their content type (text, audio, image, or document), processes them using OpenAI's language models and embeddings, and utilizes a MongoDB Atlas Vector Store for RAG to provide informed answers from a knowledge base.
What it does
- Listens for WhatsApp Messages: Triggers upon receiving any message via WhatsApp Business Cloud.
- Identifies Message Type: Determines if the incoming message is text, audio, image, or a document (like PDF).
- Processes Audio Messages: If an audio message is received, it transcribes the audio using OpenAI's Whisper API.
- Processes Image Messages: If an image message is received, it uses OpenAI's Vision API to describe the image content.
- Processes PDF/Document Messages: If a document (like a PDF) is received, it extracts text content from the file, splits it into manageable chunks, generates embeddings, and stores them in a MongoDB Atlas Vector Store for RAG.
- Processes Text Messages: For text messages, it checks if the message is a query for the RAG system or a general chat message.
- Retrieves Context with RAG: For RAG queries, it retrieves relevant information from the MongoDB Atlas Vector Store based on the user's input.
- Generates AI Responses: Utilizes an OpenAI Chat Model (GPT) to generate conversational responses, incorporating retrieved context for RAG queries.
- Maintains Conversation History: Uses a simple memory buffer to keep track of the conversation flow, enabling more coherent interactions.
- Sends WhatsApp Replies: Sends the AI-generated response back to the user via WhatsApp.
- Handles Unknown Message Types: Provides a fallback response for any message types it cannot process.
Prerequisites/Requirements
To use this workflow, you will need:
- n8n Instance: A running n8n instance.
- WhatsApp Business Cloud Account: Configured for sending and receiving messages.
- OpenAI API Key: For accessing OpenAI's language models (GPT), embeddings, Whisper (audio transcription), and Vision (image description).
- MongoDB Atlas Account: A MongoDB Atlas cluster with a Vector Search index configured for storing and retrieving document embeddings.
- Google Docs (Optional): If you intend to use Google Docs as a source for your knowledge base.
- Credentials: You will need to configure credentials for WhatsApp Business Cloud, OpenAI, and MongoDB Atlas within n8n.
Setup/Usage
- Import the workflow: Download the JSON provided and import it into your n8n instance.
- Configure WhatsApp Trigger:
- Set up your WhatsApp Business Cloud credential.
- Ensure the webhook URL is correctly configured in your WhatsApp Business Account settings to point to this n8n workflow's webhook.
- Configure OpenAI Credentials:
- Provide your OpenAI API key in the "OpenAI Chat Model", "Embeddings OpenAI", and "OpenAI" nodes.
- Configure MongoDB Atlas Vector Store:
- Set up your MongoDB Atlas credential.
- Specify your database, collection, and vector index name.
- Ensure your MongoDB Atlas cluster has a vector search index configured for the embedding field.
- Configure Google Docs (Optional):
- If you plan to ingest data from Google Docs, configure your Google Docs credential.
- Customize the Logic: Review the "Switch" node and "Code" nodes to understand how different message types are handled and customize the logic or AI prompts as needed.
- Activate the workflow: Once all credentials and configurations are set, activate the workflow to start processing WhatsApp messages.
This workflow is highly modular, allowing for easy expansion with additional tools, data loaders, and AI models to further enhance its capabilities.
Related Templates
Auto-reply & create Linear tickets from Gmail with GPT-5, gotoHuman & human review
This workflow automatically classifies every new email from your linked mailbox, drafts a personalized reply, and creates Linear tickets for bugs or feature requests. It uses a human-in-the-loop with gotoHuman and continuously improves itself by learning from approved examples. How it works The workflow triggers on every new email from your linked mailbox. Self-learning Email Classifier: an AI model categorizes the email into defined categories (e.g., Bug Report, Feature Request, Sales Opportunity, etc.). It fetches previously approved classification examples from gotoHuman to refine decisions. Self-learning Email Writer: the AI drafts a reply to the email. It learns over time by using previously approved replies from gotoHuman, with per-classification context to tailor tone and style (e.g., different style for sales vs. bug reports). Human Review in gotoHuman: review the classification and the drafted reply. Drafts can be edited or retried. Approved values are used to train the self-learning agents. Send approved Reply: the approved response is sent as a reply to the email thread. Create ticket: if the classification is Bug or Feature Request, a ticket is created by another AI agent in Linear. Human Review in gotoHuman: How to set up Most importantly, install the gotoHuman node before importing this template! (Just add the node to a blank canvas before importing) Set up credentials for gotoHuman, OpenAI, your email provider (e.g. Gmail), and Linear. In gotoHuman, select and create the pre-built review template "Support email agent" or import the ID: 6fzuCJlFYJtlu9mGYcVT. Select this template in the gotoHuman node. In the "gotoHuman: Fetch approved examples" http nodes you need to add your formId. It is the ID of the review template that you just created/imported in gotoHuman. Requirements gotoHuman (human supervision, memory for self-learning) OpenAI (classification, drafting) Gmail or your preferred email provider (for email trigger+replies) Linear (ticketing) How to customize Expand or refine the categories used by the classifier. Update the prompt to reflect your own taxonomy. Filter fetched training data from gotoHuman by reviewer so the writer adapts to their personalized tone and preferences. Add more context to the AI email writer (calendar events, FAQs, product docs) to improve reply quality.
By gotoHumanAutomate RSS to social media pipeline with AI, Airtable & GetLate for multiple platforms
Overview Automates your complete social media content pipeline: sources articles from Wallabag RSS, generates platform-specific posts with AI, creates contextual images, and publishes via GetLate API. Built with 63 nodes across two workflows to handle LinkedIn, Instagram, and Bluesky—with easy expansion to more platforms. Ideal for: Content marketers, solo creators, agencies, and community managers maintaining a consistent multi-platform presence with minimal manual effort. How It Works Two-Workflow Architecture: Content Aggregation Workflow Monitors Wallabag RSS feeds for tagged articles (to-share-linkedin, to-share-instagram, etc.) Extracts and converts content from HTML to Markdown Stores structured data in Airtable with platform assignment AI Generation & Publishing Workflow Scheduled trigger queries Airtable for unpublished content Routes to platform-specific sub-workflows (LinkedIn, Instagram, Bluesky) LLM generates optimized post text and image prompts based on custom brand parameters Optionally generates AI images and hosts them on Imgbb CDN Publishes via GetLate API (immediate or draft mode) Updates Airtable with publication status and metadata Key Features: Tag-based content routing using Wallabag's native system Swappable AI providers (Groq, OpenAI, Anthropic) Platform-specific optimization (tone, length, hashtags, CTAs) Modular design—duplicate sub-workflows to add new platforms in \~30 minutes Centralized Airtable tracking with 17 data points per post Set Up Steps Setup time: \~45-60 minutes for initial configuration Create accounts and get API keys (\~15 min) Wallabag (with RSS feeds enabled) GetLate (social media publishing) Airtable (create base with provided schema—see sticky notes) LLM provider (Groq, OpenAI, or Anthropic) Image service (Hugging Face, Fal.ai, or Stability AI) Imgbb (image hosting) Configure n8n credentials (\~10 min) Add all API keys in n8n's credential manager Detailed credential setup instructions in workflow sticky notes Set up Airtable database (\~10 min) Create "RSS Feed - Content Store" base Add 19 required fields (schema provided in workflow sticky notes) Get Airtable base ID and API key Customize brand prompts (\~15 min) Edit "Set Custom SMCG Prompt" node for each platform Define brand voice, tone, goals, audience, and image preferences Platform-specific examples provided in sticky notes Configure platform settings (\~10 min) Set GetLate account IDs for each platform Enable/disable image generation per platform Choose immediate publish vs. draft mode Adjust schedule trigger frequency Test and deploy Tag test articles in Wallabag Monitor the first few executions in draft mode Activate workflows when satisfied with the output Important: This is a proof-of-concept template. Test thoroughly with draft mode before production use. Detailed setup instructions, troubleshooting tips, and customization guidance are in the workflow's sticky notes. Technical Details 63 nodes: 9 Airtable operations, 8 HTTP requests, 7 code nodes, 3 LangChain LLM chains, 3 RSS triggers, 3 GetLate publishers Supports: Multiple LLM providers, multiple image generation services, unlimited platforms via modular architecture Tracking: 17 metadata fields per post, including publish status, applied parameters, character counts, hashtags, image URLs Prerequisites n8n instance (self-hosted or cloud) Accounts: Wallabag, GetLate, Airtable, LLM provider, image generation service, Imgbb Basic understanding of n8n workflows and credential configuration Time to customize prompts for your brand voice Detailed documentation, Airtable schema, prompt examples, and troubleshooting guides are in the workflow's sticky notes. Category Tags social-media-automation, ai-content-generation, rss-to-social, multi-platform-posting, getlate-api, airtable-database, langchain, workflow-automation, content-marketing
By Mikal Hayden-GatesAi website scraper & company intelligence
AI Website Scraper & Company Intelligence Description This workflow automates the process of transforming any website URL into a structured, intelligent company profile. It's triggered by a form, allowing a user to submit a website and choose between a "basic" or "deep" scrape. The workflow extracts key information (mission, services, contacts, SEO keywords), stores it in a structured Supabase database, and archives a full JSON backup to Google Drive. It also features a secondary AI agent that automatically finds and saves competitors for each company, building a rich, interconnected database of company intelligence. --- Quick Implementation Steps Import the Workflow: Import the provided JSON file into your n8n instance. Install Custom Community Node: You must install the community node from: https://www.npmjs.com/package/n8n-nodes-crawl-and-scrape FIRECRAWL N8N Documentation https://docs.firecrawl.dev/developer-guides/workflow-automation/n8n Install Additional Nodes: n8n-nodes-crawl-and-scrape and n8n-nodes-mcp fire crawl mcp . Set up Credentials: Create credentials in n8n for FIRE CRAWL API,Supabase, Mistral AI, and Google Drive. Configure API Key (CRITICAL): Open the Web Search tool node. Go to Parameters → Headers and replace the hardcoded Tavily AI API key with your own. Configure Supabase Nodes: Assign your Supabase credential to all Supabase nodes. Ensure table names (e.g., companies, competitors) match your schema. Configure Google Drive Nodes: Assign your Google Drive credential to the Google Drive2 and save to Google Drive1 nodes. Select the correct Folder ID. Activate Workflow: Turn on the workflow and open the Webhook URL in the “On form submission” node to access the form. --- What It Does Form Trigger Captures user input: “Website URL” and “Scraping Type” (basic or deep). Scraping Router A Switch node routes the flow: Deep Scraping → AI-based MCP Firecrawler agent. Basic Scraping → Crawlee node. Deep Scraping (Firecrawl AI Agent) Uses Firecrawl and Tavily Web Search. Extracts a detailed JSON profile: mission, services, contacts, SEO keywords, etc. Basic Scraping (Crawlee) Uses Crawl and Scrape node to collect raw text. A Mistral-based AI extractor structures the data into JSON. Data Storage Stores structured data in Supabase tables (companies, company_basicprofiles). Archives a full JSON backup to Google Drive. Automated Competitor Analysis Runs after a deep scrape. Uses Tavily web search to find competitors (e.g., from Crunchbase). Saves competitor data to Supabase, linked by company_id. --- Who's It For Sales & Marketing Teams: Enrich leads with deep company info. Market Researchers: Build structured, searchable company databases. B2B Data Providers: Automate company intelligence collection. Developers: Use as a base for RAG or enrichment pipelines. --- Requirements n8n instance (self-hosted or cloud) Supabase Account: With tables like companies, competitors, social_links, etc. Mistral AI API Key Google Drive Credentials Tavily AI API Key (Optional) Custom Nodes: n8n-nodes-crawl-and-scrape --- How It Works Flow Summary Form Trigger: Captures “Website URL” and “Scraping Type”. Switch Node: deep → MCP Firecrawler (AI Agent). basic → Crawl and Scrape node. Scraping & Extraction: Deep path: Firecrawler → JSON structure. Basic path: Crawlee → Mistral extractor → JSON. Storage: Save JSON to Supabase. Archive in Google Drive. Competitor Analysis (Deep Only): Finds competitors via Tavily. Saves to Supabase competitors table. End: Finishes with a No Operation node. --- How To Set Up Import workflow JSON. Install community nodes (especially n8n-nodes-crawl-and-scrape from npm). Configure credentials (Supabase, Mistral AI, Google Drive). Add your Tavily API key. Connect Supabase and Drive nodes properly. Fix disconnected “basic” path if needed. Activate workflow. Test via the webhook form URL. --- How To Customize Change LLMs: Swap Mistral for OpenAI or Claude. Edit Scraper Prompts: Modify system prompts in AI agent nodes. Change Extraction Schema: Update JSON Schema in extractor nodes. Fix Relational Tables: Add Items node before Supabase inserts for arrays (social links, keywords). Enhance Automation: Add email/slack notifications, or replace form trigger with a Google Sheets trigger. --- Add-ons Automated Trigger: Run on new sheet rows. Notifications: Email or Slack alerts after completion. RAG Integration: Use the Supabase database as a chatbot knowledge source. --- Use Case Examples Sales Lead Enrichment: Instantly get company + competitor data from a URL. Market Research: Collect and compare companies in a niche. B2B Database Creation: Build a proprietary company dataset. --- WORKFLOW IMAGE --- Troubleshooting Guide | Issue | Possible Cause | Solution | |-------|----------------|-----------| | Form Trigger 404 | Workflow not active | Activate the workflow | | Web Search Tool fails | Missing Tavily API key | Replace the placeholder key | | FIRECRAWLER / find competitor fails | Missing MCP node | Install n8n-nodes-mcp | | Basic scrape does nothing | Switch node path disconnected | Reconnect “basic” output | | Supabase node error | Wrong table/column names | Match schema exactly | --- Need Help or More Workflows? Want to customize this workflow for your business or integrate it with your existing tools? Our team at Digital Biz Tech can tailor it precisely to your use case from automation logic to AI-powered enhancements. Contact: shilpa.raju@digitalbiz.tech For more such offerings, visit us: https://www.digitalbiz.tech ---
By DIGITAL BIZ TECH