Build SmartCart v1 : Baseline

Business Scenario

Welcome!

Previously, you installed LangChain & GroqCloud connector, secured API key, managed dependencies in requirements.txt. You made LLM calls using ChatGroq, examined message-based request and AIMessage response, and built smartcart_v1.py as the foundation for SmartCart AI v1.

Today, your manager has assigned you a task to turn the SmartCart AI script into a working chatbot. The chatbot should keep asking what the shopper wants until they say goodbye, handle empty or unclear input, and show a simple message instead of crashing when the API has a problem

Git Pull

git pull origin branchName

Pre-Lab Preparation

Topic : Build Chatbot V1 : Baseline

1) Conversational Loop

2) Handling User Input

3) Basic Error Handling & Testing

Dataset :

Task 1: Reload Lab 9 Backend

!pip install -q -r requirements.txt 
 
import os 
from google.colab import userdata 
 
os.environ["GROQ_API_KEY"] = userdata.get("GROQ_API_KEY") 
 
from smartcart_v1 import ask_smartcart 
 
response = ask_smartcart("gym earbuds under Rs. 2,000") 
print(response.content)

Output

Task 2 : Build Conversational Loop

EXIT_WORDS = ["quit", "exit", "bye"]

print("SmartCart AI v1 -- type 'quit' to exit")

while True:
    user_input = input("\nYou: ")

    if user_input.lower() in EXIT_WORDS:
        print("SmartCart AI: Thanks for shopping with SmartCart. Bye!")
        break

    response = ask_smartcart(user_input)
    print("SmartCart AI:", response.content)

Output

Task 3 : Handle User Input

MAX_CHARS = 500   # about 120 tokens -- plenty for a shopping request

def check_input(text):
    """Return (cleaned_text, problem). problem is "" when the input is fine."""
    cleaned = " ".join(text.split())          # remove extra spaces
    if cleaned == "":
        return cleaned, "Please type what you are shopping for."
    if len(cleaned) > MAX_CHARS:
        return cleaned, f"Please keep your request under {MAX_CHARS} characters."
    return cleaned, ""

for sample in ["  Organic  Sugar Nestle  ", "     ", "cheap " * 100]:
    cleaned, problem = check_input(sample)
    print(repr(cleaned[:30]), "->", problem or "OK")

Output

Error Handling

Calling an API can fail for reasons that are not the shopper's fault. Without error handling, one failure stops the whole chatbot​

Task 4 : Add Basic Error Handling

Wrap the call in a safe function

1

FRIENDLY_ERROR = "Sorry, I'm having trouble right now. Please try again in a moment."

def safe_ask(text):
    """Return the reply text, or a friendly message if anything goes wrong."""
    try:
        response = ask_smartcart(text)
    except Exception as error:
        print("[log]", type(error).__name__, "-", str(error)[:80])
        return FRIENDLY_ERROR

    if response.response_metadata.get("finish_reason") == "length":
        return response.content + "\n(Reply was shortened.)"
    return response.content

Force an Error

2

import smartcart_v1
from langchain_groq import ChatGroq
real_llm = smartcart_v1.llm
smartcart_v1.llm = ChatGroq(model="qwen/not-a-real-model")   # break it

print("SmartCart AI:", safe_ask("gym earbuds under Rs. 2,000"))

smartcart_v1.llm = real_llm                                   # fix it

Task 5 : Complete SmartCart v1

%%writefile smartcart_v1_chat.py
from smartcart_v1 import ask_smartcart

EXIT_WORDS = ["quit", "exit", "bye"]
MAX_CHARS = 500
FRIENDLY_ERROR = "Sorry, I'm having trouble right now. Please try again in a moment."


def check_input(text):
    """Return (cleaned_text, problem). problem is "" when the input is fine."""
    cleaned = " ".join(text.split())
    if cleaned == "":
        return cleaned, "Please type what you are shopping for."
    if len(cleaned) > MAX_CHARS:
        return cleaned, f"Please keep your request under {MAX_CHARS} characters."
    return cleaned, ""


def safe_ask(text):
    """Return the reply text, or a friendly message if anything goes wrong."""
    try:
        response = ask_smartcart(text)
    except Exception as error:
        print("[log]", type(error).__name__, "-", str(error)[:80])
        return FRIENDLY_ERROR
    if response.response_metadata.get("finish_reason") == "length":
        return response.content + "\n(Reply was shortened.)"
    return response.content

%%writefile smartcart_v1_chat.py
from smartcart_v1 import ask_smartcart

EXIT_WORDS = ["quit", "exit", "bye"]
MAX_CHARS = 500
FRIENDLY_ERROR = "Sorry, I'm having trouble right now. Please try again in a moment."


def check_input(text):
    """Return (cleaned_text, problem). problem is "" when the input is fine."""
    cleaned = " ".join(text.split())
    if cleaned == "":
        return cleaned, "Please type what you are shopping for."
    if len(cleaned) > MAX_CHARS:
        return cleaned, f"Please keep your request under {MAX_CHARS} characters."
    return cleaned, ""


def safe_ask(text):
    """Return the reply text, or a friendly message if anything goes wrong."""
    try:
        response = ask_smartcart(text)
    except Exception as error:
        print("[log]", type(error).__name__, "-", str(error)[:80])
        return FRIENDLY_ERROR
    if response.response_metadata.get("finish_reason") == "length":
        return response.content + "\n(Reply was shortened.)"
    return response.content


def chat():
    """Run the chatbot until the shopper types an exit word."""
    print("SmartCart AI v1 -- type 'quit' to exit")
    while True:
        user_input = input("\nYou: ")
        if user_input.strip().lower() in EXIT_WORDS:
            print("SmartCart AI: Thanks for shopping with SmartCart. Bye!")
            break
        cleaned, problem = check_input(user_input)
        if problem:
            print("SmartCart AI:", problem)
            continue                      # skip the LLM call
        print("SmartCart AI:", safe_ask(cleaned))


if __name__ == "__main__":
    chat()

​Run this and try an empty message, a normal request, a follow-up and quit

%run smartcart_v1_chat.py

Output :

Insert your request in the box : 1) running shoes for women under 3000

                                   2) make the second one cheaper

Output :

Task 6 : Test Chatbot Behaviour

import pandas as pd
from smartcart_v1_chat import check_input, safe_ask
test_cases = [
    {"id": "T1", "type": "Normal request", "input": "gym earbuds under Rs. 2,000"},
    {"id": "T2", "type": "Hinglish",       "input": "office ke liye formal shirt chahiye under 1500 rupees"},
    {"id": "T3", "type": "Messy input",    "input": "  KIDS DRESS   GREY size 8 "},
    {"id": "T4", "type": "Empty input",    "input": "    "},
    {"id": "T5", "type": "Too long",       "input": "cheap " * 100},
    {"id": "T6", "type": "Follow-up",      "input": "make the second one cheaper"},
    {"id": "T7", "type": "Off-topic",      "input": "who won the cricket match yesterday"},
    {"id": "T8", "type": "Store policy",   "input": "What is SmartCart's return policy for opened electronics?"},]

for case in test_cases:
    cleaned, problem = check_input(case["input"])
    case["reply"] = problem if problem else safe_ask(cleaned)
    case["called_llm"] = "No" if problem else "Yes"
    print(case["id"], case["type"], "->", case["reply"][:60].replace("\n", " "))

results = pd.DataFrame(test_cases)
results.to_csv("smartcart_v1_test_results.csv", index=False)

Output

   Git Push

git push origin branchName

 

Great job!

You have successfully built a conversational loop with exit words, cleaned and validated shopper input, added basic error handling for API failures, and completed SmartCart AI v1 as a shopping assistant without memory.​

Checkpoint

Next-Lab Preparation

Topic : Prompt Engineering I : Fundamentals & Roles

1) Zero shot vs Few shot prompting

2) Instruction Clarity

3) Roleplay prompting & System prompting