docs: add simple RAG pipeline example

en/integrations/langchain/chains/conversational-retrieval-qa-chain.md

@@ -29,3 +29,60 @@ A chain for performing question-answering tasks with a retrieval component.
 | Name                           | Description                   |
 | ------------------------------ | ----------------------------- |
 | ConversationalRetrievalQAChain | Final node to return response |
+
+
+## Building a Simple RAG Pipeline
+
+This example demonstrates a simple Retrieval-Augmented Generation (RAG) pipeline using a PDF document, OpenAI embeddings, an in-memory vector store, and a Conversational Retrieval QA Chain.
+
+### Required Nodes
+
+- PDF File Loader
+- Recursive Character Text Splitter
+- OpenAI Embeddings
+- In-Memory Vector Store
+- OpenAI Chat Model
+- Conversational Retrieval QA Chain
+- Buffer Memory (Recommended)
+
+### Node Connections
+
+```text
+Recursive Character Text Splitter → PDF File Loader
+
+PDF File Loader → In-Memory Vector Store
+
+OpenAI Embeddings → In-Memory Vector Store
+
+In-Memory Vector Store (Retriever) → Conversational Retrieval QA Chain
+
+OpenAI Chat Model → Conversational Retrieval QA Chain
+
+Buffer Memory → Conversational Retrieval QA Chain
+```
+
+### Explanation
+
+The PDF File Loader loads the document content, while the Recursive Character Text Splitter breaks the document into smaller chunks for retrieval.
+
+The OpenAI Embeddings node converts document chunks into vector representations. These embeddings are then stored in the In-Memory Vector Store together with the document chunks.
+
+When a user asks a question, the Conversational Retrieval QA Chain retrieves the most relevant document chunks from the vector store and sends them to the chat model to generate a grounded response.
+
+### Common Beginner Mistakes
+
+- Connecting the embedding model directly to the Conversational Retrieval QA Chain
+- Forgetting to connect the vector store retriever to the Conversational Retrieval QA Chain
+- Confusing embeddings with chat models
+- Forgetting to include required prompt variables such as `{context}` and `{question}`
+- Connecting incorrect node output types together
+- Forgetting to add a Memory node, which prevents the chain from remembering previous interactions
+
+### Recommended Prompt Rules
+
+To reduce hallucinations and unsupported responses, consider adding rules such as:
+
+- Use only the provided context to answer
+- Do not use external knowledge
+- If the information is not available in the provided context, clearly state that it was not found
+- Avoid unsupported assumptions