Update app.py

app.py

@@ -1,88 +1,110 @@
-
 import os
 import groq
-import phi
-import phi.api
-
 from dotenv import load_dotenv
-from phi.agent import Agent
-from phi.model.groq import Groq
-from phi.tools.yfinance import YFinanceTools
-from phi.tools.youtube_tools import YouTubeTools
-from phi.tools.googlesearch import GoogleSearch
-from phi.playground import Playground
+import gradio as gr
+
+
+# 1. Environment and Groq Client Setup
+
+load_dotenv()  # Load environment variables from .env
+
+# Retrieve the API key and model identifier from environment variables
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+GROQ_MODEL_ID = os.getenv("GROQ_MODEL_ID", "llama-3.3-70b-versatile")
 
-# 1. Load environment variables
-load_dotenv()
+# Initialize the Groq client (adjust this based on your actual Groq API)
+groq_client = groq.Client(api_key=GROQ_API_KEY)
 
-phi.api = os.getenv("PHI_API_KEY")
-groq_api_key = os.getenv("GROQ_API_KEY")
 
-# 2. Create a groq client
-groq_client = groq.Client(api_key=groq_api_key)
+# 2. Define a Simple Agent Class Using Groq
 
-# 3. Define Agents
+class Agent:
+    def __init__(self, name, role, instructions):
+        self.name = name
+        self.role = role
+        self.instructions = instructions
+
+    def build_prompt(self, query):
+        # Construct a prompt that provides context for the agent
+        prompt = (
+            f"Agent Name: {self.name}\n"
+            f"Role: {self.role}\n"
+            f"Instructions: {self.instructions}\n"
+            f"User Query: {query}\n"
+            "Response:"
+        )
+        return prompt
+
+    def respond(self, query):
+        prompt = self.build_prompt(query)
+        # Call Groq Cloud API to generate a response (adjust parameters as needed)
+        response = groq_client.generate(
+            model_id=GROQ_MODEL_ID,
+            prompt=prompt,
+            max_tokens=200
+        )
+        # Assume the API returns a dict with a 'text' field for the generated response
+        return response.get("text", "")
+
+
+# 3. Define Specific Agents
 
-# Websearch Agent
 websearch_agent = Agent(
-    name='websearch_agent',
-    role="Search the web for financial information including all the latest news too.",
-    model=Groq(id="llama-3.3-70b-versatile"),
-    tools=[GoogleSearch()],
-    instructions=["Always include the sources for the information in APA format."],
-    markdown=True,
-    # debug_mode=True,
+    name="WebSearch Agent",
+    role="Searches the web for financial information and the latest news.",
+    instructions="Return results with sources in APA format."
 )
 
-# Youtube Agent
 youtube_agent = Agent(
-    name="YouTube Agent", 
-    role="You are a YouTube agent that has the special skill of understanding YouTube videos and answering questions about them. You can only answer if the duration is less than 22 minutes.",
-    model=Groq(id="llama-3.3-70b-versatile"),
-    tools=[YouTubeTools()],
-    instructions=[
-        "1. When the user asks about a video, confirm that they have provided a valid YouTube URL. If not, ask them for it.",
-        "2. Using a video URL, get the video data using the get_youtube_video_data tool. Using the video data, get the video captions using the get_youtube_video_captions tool.",
-        "3. Using the data and captions, answer the user questions in an engaging and thoughtful manner and only focus on the important information.",
-        "4. If you cannot find the information, let the user know by asking for more details, and don't hallucinate.",
-        "5. Keep your answers concise and engaging."
-    ],
-    markdown=True,
-    read_chat_history=True,
-    # debug_mode=True,
+    name="YouTube Agent",
+    role="Analyzes YouTube videos that are less than 22 minutes in duration.",
+    instructions="Provide concise and engaging answers based on the video content."
 )
 
-# Financial Agent
 finance_agent = Agent(
-    name="Finance AI Agent",
-    model=Groq(id="llama-3.3-70b-versatile"),
-    tools=[
-        YFinanceTools(
-            stock_price=True,
-            analyst_recommendations=True,
-            stock_fundamentals=True,
-            company_news=True
-        ),
-    ],
-    description="You are an investment analyst that researches stock prices, analyst recommendations, and stock fundamentals.",
-    instructions=["Format your response using markdown and use tables to display data where possible."],
-    markdown=True,
-    # debug_mode=True,
+    name="Finance Agent",
+    role="Analyzes stock prices, analyst recommendations, and fundamentals.",
+    instructions="Format your response using markdown tables where applicable."
 )
 
-# Multi-Model Agent
-MultiModelAgent = Agent(
-    team=[websearch_agent, finance_agent, youtube_agent],
-    model=Groq(id="llama-3.3-70b-versatile"),
-    name="Multi-Model Agent",
-    instructions=["Always include sources and use tables to display data where possible."],
-    show_tool_calls=True,
-    markdown=True,
-    # debug_mode=True,
-)
 
-# 4. Create the Playground app with all agents
-playground = Playground(agents=[websearch_agent, finance_agent, MultiModelAgent, youtube_agent])
+# 4. Build a Multi-Agent Coordinator
+
+class MultiAgent:
+    def __init__(self, agents):
+        self.agents = agents
+
+    def handle_query(self, query):
+        # For the given query, ask each agent and collect their responses
+        responses = {}
+        for agent in self.agents:
+            responses[agent.name] = agent.respond(query)
+        return responses
+
+# Create a multi-agent system with the defined agents
+multi_agent = MultiAgent([websearch_agent, youtube_agent, finance_agent])
+
+
+# 5. Gradio Chatbot Front End Function
+
+def chat_response(user_query):
+    responses = multi_agent.handle_query(user_query)
+    # Format the responses into a single markdown string for display
+    result = ""
+    for agent_name, response in responses.items():
+         result += f"**{agent_name}**:\n{response}\n\n"
+    return result
+
+
+# 6. Create the Gradio Interface
+
+demo = gr.Interface(
+    fn=chat_response,
+    inputs=gr.Textbox(placeholder="Enter your message here...", lines=2),
+    outputs="markdown",
+    title="Groq Agent Chatbot",
+    description="A chatbot powered by Groq Cloud for GPT inference with multiple agents."
+)
 
-# 5. The Hugging Face Space will look for 'app' to serve
-app = playground.get_app()
+if __name__ == "__main__":
+    demo.launch()