Add application file

.gitignore

@@ -0,0 +1,2 @@
+.env
+backend/.env

API_TEST_RESULTS.txt

@@ -0,0 +1,80 @@
+## ✅ API Endpoint Test Results
+
+### 1. Root Endpoint
+**Request:**
+```bash
+curl.exe -X GET "http://localhost:8000/"
+```
+
+**Response:**
+```json
+{"message":"RAG Chatbot API"}
+```
+**Status:** ✅ PASS
+
+---
+
+### 2. Health Check Endpoint
+**Request:**
+```bash
+curl.exe -X GET "http://localhost:8000/api/health"
+```
+
+**Response:**
+```json
+{"status":"ok"}
+```
+**Status:** ✅ PASS
+
+---
+
+### 3. Chat Endpoint (RAG-Powered)
+**Request:**
+```bash
+curl.exe -X POST "http://localhost:8000/api/chat" \
+  -H "Content-Type: application/json" \
+  --data "@test_request.json"
+```
+
+**Request Body** (`test_request.json`):
+```json
+{
+  "question": "What is RAG?",
+  "user_id": 1
+}
+```
+
+**Response:**
+```json
+{
+  "answer": "RAG, or Retrieval-Augmented Generation, is a machine learning approach that combines retrieval-based techniques with generative models, particularly in the context of natural language processing (NLP). The main idea behind RAG is to enhance the capabilities of generative models (like language models) by integrating them with external knowledge sources or databases.\n\nIn RAG, when a model receives a prompt or query, it first retrieves relevant documents or information from a knowledge base using a retrieval mechanism. Then, it uses this retrieved information to inform and augment its generative response, effectively producing more accurate and contextually relevant answers. This approach allows the model to leverage both broad generative capabilities and specific, factual knowledge, leading to improved performance in tasks like question answering, summarization, and conversational agents.",
+  "sources": []
+}
+```
+**Status:** ✅ PASS
+
+**Note:** Sources array is empty because no documents have been ingested yet. To populate the vector database, run:
+```bash
+python scripts/ingest_content.py
+```
+
+---
+
+## 🎯 All API Endpoints Working!
+
+### Backend Configuration:
+- **OpenAI API:** ✅ Connected (using gpt-4o)
+- **Database:** ✅ Connected (Neon Postgres)
+- **Qdrant:** ✅ Connected (Qdrant Cloud)
+- **Server:** ✅ Running on http://localhost:8000
+
+### API Documentation:
+Visit http://localhost:8000/docs for interactive API documentation (Swagger UI)
+
+---
+
+## Next Steps:
+1. ✅ Backend is fully operational
+2. 📝 Ingest documentation content (optional): `python scripts/ingest_content.py`
+3. 🚀 Start frontend: `cd physical-ai-humanoid-robotics && npm start`
+4. 🧪 Test chat widget on http://localhost:3000

app/services/rag_service.py

@@ -2,7 +2,6 @@
 import os
 import asyncio
 from qdrant_client import QdrantClient
-from qdrant_client.models import NamedVector
 from typing import List
 
 from app.services.openai_service import OpenAIService
@@ -18,16 +17,16 @@ class RAGService:
     async def retrieve_context(self, query: str, top_k: int = 3) -> List[str]:
         query_vector = await self.embeddings_service.create_embedding(query)
         
-        # Run synchronous Qdrant query in thread pool
+        # Use search method - compatible with all Qdrant versions
         search_result = await asyncio.to_thread(
-            self.qdrant_client.query_points,
+            self.qdrant_client.search,
             collection_name=self.collection_name,
-            query=query_vector,
+            query_vector=query_vector,
             limit=top_k,
             with_payload=True
         )
 
-        context = [point.payload.get("content", "") for point in search_result.points if point.payload]
+        context = [point.payload.get("content", "") for point in search_result if point.payload]
         return context
 
     async def generate_response(self, query: str, context: List[str]) -> str: