Update app.py

app.py

@@ -1,5 +1,4 @@
 from flask import Flask, render_template, request, jsonify, send_file
-import ollama
 import json
 import re
 from duckduckgo_search import DDGS
@@ -10,6 +9,8 @@ import urllib3
 import pandas as pd
 import io
 import ast
+from groq import Groq
+import os
 
 
 app = Flask(__name__)
@@ -99,18 +100,45 @@ Reference json:
 }
 """
 
-def ask_ollama(user_message, model='gemma3:1b', system_prompt=search_prompt):
-    response = ollama.chat(model=model, messages=[
-        {
-            "role": "system",
-            "content": system_prompt
-        },
-        {
-            "role": "user", 
-            "content": user_message
-        }
-    ])
-    ai_reply = response['message']['content']
+def ask_llm(user_message, model='llama-3.3-70b-versatile', system_prompt="You are a helpful assistant."):
+    client = Groq(api_key=os.environ.get("GROQ_API_KEY"))
+
+    response = client.chat.completions.create(
+        model=model,
+        messages=[
+            {
+                "role": "system",
+                "content": system_prompt
+            },
+            {
+                "role": "user", 
+                "content": user_message
+            }
+        ],
+        stream=False,
+    )
+
+    return response.choices[0].message.content
+
+def ask_ollama(user_message, model='llama-3.3-70b-versatile', system_prompt=search_prompt):
+    client = Groq(api_key=os.environ.get("GROQ_API_KEY"))
+
+    response = client.chat.completions.create(
+        model=model,
+        messages=[
+            {
+                "role": "system",
+                "content": system_prompt
+            },
+            {
+                "role": "user", 
+                "content": user_message
+            }
+        ],
+        stream=False,
+    )
+    
+    ai_reply = response.choices[0].message.content
     print(f"AI REPLY json:\n{ai_reply}")
 
     # Process the response to ensure we return valid JSON
@@ -118,13 +146,14 @@ def ask_ollama(user_message, model='gemma3:1b', system_prompt=search_prompt):
         # First, try to parse it directly in case it's already valid JSON
         print(f"AI REPLY:\n{ai_reply}")
         return ast.literal_eval(ai_reply.replace('json\n', '').replace('```', ''))
-    except:
+    except Exception as e:
         print(f"ERROR:\n{e}")
         # If it's not valid JSON, try to extract JSON from the text
         return {
             "1": "Error parsing response. Please try again.",
             "2": "Error parsing response. Please try again."
         }
+
 def search_web(topic, max_references=5, data_type="pdf"):
     """Search the web using DuckDuckGo and return results."""
     doc_list = []
@@ -538,19 +567,7 @@ Example: ["Insight 1", "Insight 3", "Insight 5"]"""
 ALL AVAILABLE INSIGHTS:
 {json.dumps(insights)}"""
         
-        # Call Ollama with the message and prompt
-        response = ollama.chat(model='gemma3:1b', messages=[
-            {
-                "role": "system",
-                "content": ai_select_prompt
-            },
-            {
-                "role": "user", 
-                "content": message
-            }
-        ])
-        
-        ai_reply = response['message']['content']
+        ai_reply = ask_llm(message, system_prompt=ai_select_prompt)
         print(f"AI SELECTION RESPONSE: {ai_reply}")
         
         # Parse the JSON array from the response
@@ -606,4 +623,4 @@ def generate_key_issues():
         return jsonify({'error': str(e), 'key_issues': []})
 
 if __name__ == '__main__':
-    app.run(debug=True)
+    app.run(host="0.0.0.0", port=7860)