Update app.py

app.py

@@ -1,44 +1,47 @@
 import os
-from flask import Flask, render_template, request, jsonify
-import chromadb
+from flask import Flask, request, jsonify
 from huggingface_hub import InferenceClient
-from sentence_transformers import SentenceTransformer
+import requests
 
 app = Flask(__name__)
 
-client = InferenceClient(model="meta-llama/Llama-3.2-3B-Instruct", token=os.getenv("HF_TOKEN"))
-
-embed_model = SentenceTransformer("all-MiniLM-L6-v2")
-db_client = chromadb.PersistentClient(path="./my_vector_db")
-collection = db_client.get_or_create_collection(name="research_docs")
-
-def add_to_knowledge_base(text_list):
-    embeddings = embed_model.encode(text_list).tolist()
-    ids = [f"id_{i}" for i in range(len(text_list))]
-    collection.add(documents=text_list, embeddings=embeddings, ids=ids)
-
-initial_data = [
-    "The 2026 World AI Summit took place in Tokyo, focusing on decentralized models.",
-    "Nexus AI Research OS was launched in January 2026 as a browser-based tool.",
-]
-add_to_knowledge_base(initial_data)
-
-@app.route('/')
-def index():
-    return render_template('index.html')
+# 1. Setup Clients
+hf_client = InferenceClient(model="meta-llama/Llama-3.2-3B-Instruct", token=os.getenv("HF_TOKEN"))
+TAVILY_API_KEY = os.getenv("TAVILY_API_KEY")
+
+def search_tavily(query):
+    """Someone else's free API doing the hard work for you."""
+    url = "https://api.tavily.com/search"
+    payload = {
+        "api_key": TAVILY_API_KEY,
+        "query": query,
+        "search_depth": "basic",
+        "max_results": 3
+    }
+    response = requests.post(url, json=payload)
+    results = response.json().get("results", [])
+    # Join snippets into one string
+    return "\n".join([f"Source: {r['url']}\nContent: {r['content']}" for r in results])
 
 @app.route('/ask', methods=['POST'])
 def ask():
     user_query = request.get_json().get("query")
     
-    query_vec = embed_model.encode([user_query]).tolist()
-    results = collection.query(query_embeddings=query_vec, n_results=2)
-    retrieved_text = "\n".join(results['documents'][0])
+    # STEP 1: Get real-time data from Tavily API
+    web_data = search_tavily(user_query)
 
-    prompt = f"Use this data to answer: {retrieved_text}\n\nQuestion: {user_query}"
+    # STEP 2: Use Llama 3.2 to synthesize the answer
+    system_prompt = f"""You are a 2026 AI Researcher. 
+    Use the following LIVE WEB DATA to answer. 
+    If the data is about the PM of India or Super Bowl, use it!
     
-    response = client.chat_completion(
-        messages=[{"role": "user", "content": prompt}],
+    DATA: {web_data}"""
+
+    response = hf_client.chat_completion(
+        messages=[
+            {"role": "system", "content": system_prompt},
+            {"role": "user", "content": user_query}
+        ],
         max_tokens=500
     )