"Financial agent code commit"

Budgeting/Agent.py

@@ -0,0 +1,25 @@
+# budgeting/agent.py
+from langchain_core.messages import BaseMessage
+from langchain_huggingface import HuggingFaceEndpoint
+
+class BudgetingAgent:
+    def __init__(self, model_id="meta-llama/Llama-3.2-3B-Instruct"):
+        self.model = HuggingFaceEndpoint(
+            repo_id=model_id,
+            task="text_generation",
+            max_new_tokens=512,
+            temperature=0.3,
+            streaming=True,
+        )
+
+    def process_budgeting_query(self, state):
+        """
+        Process the user query related to budgeting.
+        Args:
+            state (dict): The current state containing user messages.
+        """
+        messages = state["messages"]
+        question = messages[0].content
+
+        response = self.model.invoke([{"content": question}])
+        return {"messages": [response]}

Budgeting/_init_.py

@@ -0,0 +1,4 @@
+# budgeting/__init__.py
+from .agent import BudgetingAgent
+from .prompt import budgeting_prompt
+from .workflow import budgeting_workflow

Budgeting/budgeting_workflow.py

@@ -0,0 +1,19 @@
+# Budgeting/budgeting_workflow.py
+from prefect import task, flow
+
+@task
+def fetch_budget_data():
+    # Placeholder for fetching budget data
+    return {"budget": 5000, "expenses": 2000, "savings": 3000}
+
+@task
+def calculate_savings(data):
+    savings = data["budget"] - data["expenses"]
+    return savings
+
+@flow
+def budgeting_workflow(state):
+    # This flow orchestrates the budgeting process
+    budget_data = fetch_budget_data()
+    savings = calculate_savings(budget_data)
+    return {"messages": [{"content": f"Savings: {savings}"}]}

Budgeting/prompt.py

@@ -0,0 +1,9 @@
+# budgeting/prompt.py
+from langchain.prompts import PromptTemplate
+
+budgeting_prompt = PromptTemplate(
+    template="""You are a financial advisor helping the user with budgeting. \n
+    Here is the user's query: {question}\n
+    Provide actionable advice on how they can manage their budget effectively.""",
+    input_variables=["question"],
+)

Credit/Agent.py

@@ -0,0 +1,24 @@
+# credit/agent.py
+from langchain_huggingface import HuggingFaceEndpoint
+
+class CreditAgent:
+    def __init__(self, model_id="meta-llama/Llama-3.2-3B-Instruct"):
+        self.model = HuggingFaceEndpoint(
+            repo_id=model_id,
+            task="text_generation",
+            max_new_tokens=512,
+            temperature=0.3,
+            streaming=True,
+        )
+
+    def process_credit_query(self, state):
+        """
+        Process the user query related to credit.
+        Args:
+            state (dict): The current state containing user messages.
+        """
+        messages = state["messages"]
+        question = messages[0].content
+
+        response = self.model.invoke([{"content": question}])
+        return {"messages": [response]}

Credit/Prompt.py

@@ -0,0 +1,9 @@
+# credit/prompt.py
+from langchain.prompts import PromptTemplate
+
+credit_prompt = PromptTemplate(
+    template="""You are a credit advisor helping the user understand credit management. \n
+    Here is the user's query: {question}\n
+    Provide detailed guidance on credit management and options for the user.""",
+    input_variables=["question"],
+)

Credit/_init_.py

@@ -0,0 +1,4 @@
+# credit/__init__.py
+from .agent import CreditAgent
+from .prompt import credit_prompt
+from .workflow import credit_workflow

Credit/credit_workflow.py

@@ -0,0 +1,22 @@
+# Credit/credit_workflow.py
+from prefect import task, flow
+
+@task
+def fetch_credit_data(state):
+    # Simulated credit data
+    return {"credit_id": state, "limit": 5000, "used": 1200}
+
+@task
+def calculate_remaining_credit(data):
+    # Calculate remaining credit
+    limit = data["limit"]
+    used = data["used"]
+    remaining_credit = limit - used
+    return remaining_credit
+
+@flow
+def credit_workflow(state):
+    credit_data = fetch_credit_data(state)
+    remaining_credit = calculate_remaining_credit(credit_data)
+    # Return in the expected format
+    return {"messages": [{"content": f"Remaining credit amount is: {remaining_credit}"}]}

Loan/Agent.py

@@ -0,0 +1,24 @@
+# loan/agent.py
+from langchain_huggingface import HuggingFaceEndpoint
+
+class LoanAgent:
+    def __init__(self, model_id="meta-llama/Llama-3.2-3B-Instruct"):
+        self.model = HuggingFaceEndpoint(
+            repo_id=model_id,
+            task="text_generation",
+            max_new_tokens=512,
+            temperature=0.3,
+            streaming=True,
+        )
+
+    def process_loan_query(self, state):
+        """
+        Process the user query related to loans.
+        Args:
+            state (dict): The current state containing user messages.
+        """
+        messages = state["messages"]
+        question = messages[0].content
+
+        response = self.model.invoke([{"content": question}])
+        return {"messages": [response]}

Loan/Financial-Agent.code-workspace

@@ -0,0 +1,8 @@
+{
+	"folders": [
+		{
+			"path": ".."
+		}
+	],
+	"settings": {}
+}

Loan/Prompt.py

@@ -0,0 +1,9 @@
+# loan/prompt.py
+from langchain.prompts import PromptTemplate
+
+loan_prompt = PromptTemplate(
+    template="""You are a loan advisor assisting the user with their loan inquiries. \n
+    Here is the user's query: {question}\n
+    Provide a detailed response regarding the loan options, eligibility, and best advice.""",
+    input_variables=["question"],
+)

Loan/_init_.py

@@ -0,0 +1,4 @@
+# loan/__init__.py
+from .agent import LoanAgent
+from .prompt import loan_prompt
+from .workflow import loan_workflow

Loan/loan_workflow.py

@@ -0,0 +1,32 @@
+# Loan/loan_workflow.py
+
+from prefect import task, flow
+
+# Define a Prefect task for loan data processing
+@task
+def fetch_loan_data(loan_id):
+    # Placeholder for fetching loan data (could be from a database or API)
+    loan_data = {"loan_id": loan_id, "amount": 10000, "interest_rate": 5.5}
+    #print(f"Fetched loan data: {loan_data}")
+    return loan_data
+
+# Define a task to calculate loan repayment
+@task
+def calculate_repayment(loan_data):
+    amount = loan_data["amount"]
+    interest_rate = loan_data["interest_rate"]
+    repayment = amount * (1 + interest_rate / 100)
+    #print(f"Calculated repayment amount: {repayment}")
+    return repayment
+
+# Define a Prefect flow to tie tasks together
+@flow
+def loan_workflow(loan_id):
+    loan_data = fetch_loan_data(loan_id)
+    repayment = calculate_repayment(loan_data)
+    #print(f"Final repayment for loan {loan_id}: {repayment}")
+    return {"messages": [{"content": f"Loan repayment amount is: {repayment}"}]}
+
+# Call the workflow in main block
+if __name__ == "__main__":
+    loan_workflow(12345)  # Pass a sample loan ID

budget_report.txt

@@ -0,0 +1,5 @@
+
+    Budget Report:
+    --------------
+    Remaining Budget: 1000
+

main.py

@@ -0,0 +1,37 @@
+# main.py
+import logging
+from Budgeting.budgeting_workflow import budgeting_workflow
+from Loan.loan_workflow import loan_workflow
+from Credit.credit_workflow import credit_workflow
+
+logging.getLogger("prefect").setLevel(logging.CRITICAL)
+
+def handle_query(query):
+    state = {"messages": [{"content": query}]}
+
+    if "budget" in query.lower():
+        response = budgeting_workflow(state)  # Call the function here
+    elif "loan" in query.lower():
+        response = loan_workflow(state)
+    elif "credit" in query.lower():
+        response = credit_workflow(state)
+    else:
+        response = {"messages": [{"content": "I'm not sure about that."}]}
+
+    return response
+
+def main():
+    print("Welcome to the Financial Agent Chatbot! ")
+    while True:
+        user_input = input("You: ")
+
+        if user_input.lower() in ["exit", "quit"]:
+            print("Bot: Goodbye! Have a great day!")
+            break
+
+        response = handle_query(user_input)
+        print(f"Bot: {response['messages'][0]['content']}")
+
+
+if __name__ == "__main__":
+    main()