Skip to main content

Recommending a Product

🛍️ Quickstart Guide - Recommending a Product​

Learn how to integrate CrewAI with Fiscus SDK and Crew in under 5 minutes to recommend products using an AI-powered chatbot.

How to Use the Fiscus SDK​

Turn natural language into powerful API operations in minutes. Follow these 5 steps to create an AI-powered product recommendation chatbot using the Fiscus SDK and Crew.

What you’ll achieve: By the end of this guide, you’ll have an AI agent capable of recommending a product to a customer through a chatbot.

1. Install Fiscus SDK​

Install Packages

pip install fiscus crewai

2. Connect Your Chatbot API (e.g., Twilio, Intercom)​

Quickly integrate with a chatbot API

from fiscus import FiscusClient, FiscusUser

# Initialize the Fiscus client with your API key and user_id
client = FiscusClient(api_key='YOUR_FISCUS_API_KEY', user_id='user_123')

# Add the chatbot connector (e.g., Intercom)
client.user.add_connector('Intercom')

# If the connector requires authentication, this will trigger an authentication flow
client.user.authenticate_connector('Intercom')

3. Initialize OpenAI​

Set up your LLM or Agent.

from openai import OpenAI

# Initialize OpenAI
language_model = OpenAI(api_key='YOUR_OPENAI_API_KEY')

4. Define a Task and Generate a Product Recommendation​

Using natural language, let’s ask OpenAI to recommend a product.

from crewai import Agent, Task, Crew

# Define a custom Fiscus Agent that uses OpenAI to handle user input
class FiscusAgent(Agent):
    def execute(self, task):
        user_input = task.description
        # Use Fiscus SDK's ai_execute to process the natural language input
        response = client.ai_execute(input=user_input, user=user)
        return response.result

# Initialize the agent
agent = FiscusAgent(name='FiscusAgent')

# Define a task asking for a product recommendation
task = Task(
    description='Recommend a product for a customer who is interested in fitness.',
    agent=agent
)

# Initialize Crew and run the task to get the recommendation
crew = Crew(agents=[agent], tasks=[task])
result = crew.run()

print(result)

5. Send the Recommendation to the Customer​

Now, let’s send the recommendation back to the customer via the chatbot:

# Assuming the AI response contains the product recommendation
recommended_product = result

# Execute the action using the Fiscus SDK to send the recommendation
send_result = client.execute(
    connector_name='Intercom', 
    operation='send_message', 
    params={
        'recipient_id': 'customer_123',
        'message': f'Recommended product: {recommended_product}'
    }
)

print(send_result)s

 

Look at That! 🎉

Success! You’ve now:

🔐 Authenticated your chatbot API account with Fiscus SDK
🤖 Used OpenAI to generate a personalized product recommendation
📤 Sent a product recommendation to your customer through the chatbot
✅ Successfully executed the recommendation flow using AI and Fiscus SDK