{"id":8189,"date":"2023-11-22T17:27:55","date_gmt":"2023-11-23T01:27:55","guid":{"rendered":"https:\/\/live-cometml.pantheonsite.io\/?p=8189"},"modified":"2025-04-24T17:04:22","modified_gmt":"2025-04-24T17:04:22","slug":"create-a-simple-e-commerce-chatbot-with-openai","status":"publish","type":"post","link":"https:\/\/www.comet.com\/site\/blog\/create-a-simple-e-commerce-chatbot-with-openai\/","title":{"rendered":"Create a Simple E-commerce Chatbot With OpenAI"},"content":{"rendered":"\n

Forget about complicated Deep Learning algorithms \u2014 making a chatbot is way simpler with OpenAI and CometLLM<\/h2>\n\n\n\n
\"Create
Chatbot in Ecommerce (Credit: Maksym Pundyk in ideainyou.com)<\/figcaption><\/figure>\n\n\n\n
\n
\n
\n

Table of Contents<\/span><\/h3>\n

Introduction<\/em><\/p>\n

I. Prompting for Developers: A Guideline<\/em><\/p>\n

II. Building a Shopbot using OpenAI and CometLLM<\/em><\/p>\n

    \n
  1. Understanding the Basics<\/em><\/li>\n
  2. Setting Up the Environment<\/em><\/li>\n
  3. Defining Product Information<\/em><\/li>\n
  4. Creating the Chatbot<\/em><\/li>\n
  5. Interacting with Customers<\/em><\/li>\n<\/ol>\n

    III. Reference<\/em><\/p>\n

    Conclusion<\/em><\/p>\n

    Introduction<\/h3>\n

    OpenAI makes it easier than ever to create chatbots for your e-commerce business, eliminating the need for complex deep learning algorithms. Discover simple and effective approaches to increase customer engagement and optimize the online shopping experience. This step-by-step tutorial will walk you through the key components and make it easy to bring your own AI-powered ShopBot to life. Easily transform your e-commerce platform into an interactive, customer-centric shopping destination.<\/p>\n

    \n
    \n
    \n<\/picture>\n<\/div>\n<\/div>\n<\/figure>\n

    \ud83d\udca1I write about Machine Learning on Medium<\/a> || Github<\/a> || Kaggle<\/a> || Linkedin<\/a>. \ud83d\udd14 Follow \u201cNhi Yen\u201d for future updates!<\/p><\/blockquote>\n

    \ud83d\udc49 Get UNLIMITED access to every story on Medium with just $1\/week \u2014 <\/em>HERE<\/em><\/strong><\/a><\/p>\n

    I. Prompting for Developers: A Guideline<\/h3>\n

    Principle 1: Write Clear and Specific Instructions<\/h4>\n
      \n
    1. Delimiters for Precision<\/strong>: start clearly. Use delimiters such as triple quotes (\u201c\u2018xxx\u2019\u201d<\/code>), triple backticks (```<\/code>), triple dashes ( \u2014 -<\/code>), angle brackets (< ><\/code>), and XML tags. These act as navigational aids, precisely guiding the model through the various components of the message.<\/li>\n
    2. Structured output for increased accuracy<\/strong>: Specify the desired format of the output (HTML or JSON). A structured framework ensures that the model accurately understands and fulfills requests, avoiding misunderstandings.<\/li>\n
    3. Validate trust assumptions<\/strong>: Validate whether the required conditions are met before assigning a task. Clearly explain the prerequisites required for the task and ensure that your understanding of the model aligns with your expectations for safe execution.<\/li>\n
    4. Few-Shot Prompting for Guidance<\/strong>: Before instructing the model, provide examples of successful task completion. This not only guides the model on the right path, but also increases the probability of accurate results.<\/li>\n<\/ol>\n

      Principle 2: Give the Model Time to \u201cThink\u201d<\/h4>\n
        \n
      1. Comprehensive task steps<\/strong>: Break down complex tasks into step-by-step instructions. This allows the model to understand the complexity of the task and increases the likelihood of accurate and complete results.<\/li>\n
      2. Encourage thoughtful solutions<\/strong>: Give it time to find its own solutions instead of urging the model\u2019s conclusions. This patient approach reduces the potential for error caused by hasty action and encourages more thoughtful responses., enhancing the likelihood of accurate and comprehensive outputs.<\/li>\n<\/ol>\n

        A Complex prompt<\/h3>\n

        Now, considering the concept of complex indicators, these principles can be applied to indicators that cover at least 3 of the following parameters:<\/p>\n

          \n
        1. Role: Assign an expert role to a task.<\/li>\n
        2. Task: Enter a brief summary of the task.<\/li>\n
        3. Context: Provide relevant background.<\/li>\n
        4. Steps to complete the task: Provide specific steps.<\/li>\n
        5. Output format: Specify the desired format, style, or tone.<\/li>\n
        6. Limitations: Please note any limitations or requirements.<\/li>\n
        7. Purpose: Describe the purpose for the results.<\/li>\n<\/ol>\n

          Example:<\/p>\n

          You<\/span> are<\/span> a<\/span> travel<\/span> assistant.<\/span>\n\nPlease<\/span> plan<\/span> a<\/span> week-long<\/span> itinerary<\/span> for<\/span> a<\/span> solo<\/span> traveler<\/span> visiting<\/span> Paris.<\/span>\n\nThe<\/span> traveler<\/span> is<\/span> interested<\/span> in<\/span> art,<\/span> history,<\/span> and<\/span> local<\/span> cuisine.<\/span> They<\/span> prefer<\/span> walking<\/span> and<\/span> public<\/span> transportation.<\/span>\n\nFollow the steps below:<\/span>\n-<\/span> Provide<\/span> a<\/span> list<\/span> of<\/span> must-visit<\/span> art<\/span> museums<\/span> and<\/span> historical<\/span> sites.<\/span>\n-<\/span> Suggest<\/span> walking<\/span> routes<\/span> that<\/span> connect<\/span> these<\/span> attractions<\/span> efficiently.<\/span>\n-<\/span> Recommend<\/span> local<\/span> restaurants<\/span> for<\/span> each<\/span> day,<\/span> considering<\/span> the<\/span> traveler's<\/span> preference<\/span> for<\/span> diverse<\/span> cuisine.<\/span>\n\nPresent<\/span> the<\/span> itinerary<\/span> in<\/span> a<\/span> clear<\/span> and<\/span> organized<\/span> document,<\/span> with<\/span> each<\/span> day's<\/span> plan<\/span> distinctively<\/span> outlined.<\/span>\n\nKeep<\/span> the<\/span> budget<\/span> moderate,<\/span> and<\/span> consider<\/span> any<\/span> current<\/span> travel<\/span> restrictions<\/span> or<\/span> events<\/span> in<\/span> Paris.<\/span>\n\nCreate<\/span> an<\/span> engaging<\/span> and<\/span> culturally<\/span> rich<\/span> itinerary<\/span> that<\/span> balances<\/span> art,<\/span> history,<\/span> and<\/span> local<\/span> culinary<\/span> experiences<\/span> for<\/span> the<\/span> solo<\/span> traveler.<\/span><\/span><\/pre>\n<\/div>\n<\/div>\n<\/div>\n\n\n\n
          <\/div>\n\n\n\n
          \n
          \n
          \n
          Want to learn how to build modern software with LLMs using the newest tools and techniques in the field? <\/em>Check out this free LLMOps course<\/em><\/a> from industry expert Elvis Saravia of DAIR.AI!<\/em><\/p><\/blockquote>\n<\/div>\n<\/div>\n<\/div>\n\n\n\n
          <\/div>\n\n\n\n
          \n
          \n
          \n
          II. Building a Shopbot using OpenAI and CometLLM<\/h2>\n
          1. Understanding the Basics<\/h3>\n
          Before we dive into the code, let\u2019s understand the main components and features of a chatbot. This chatbot, called ShopBot, helps customers find products, provides information, and guides them through the checkout process. It is friendly and helpful and can recommend products based on customer preferences.<\/p>\n
          \ud83d\udc49 Feel free to check out this <\/em>GitHub repository<\/em><\/strong><\/a> as we move forward in the process.<\/em><\/p>\n
          2. Setting Up the Environment<\/h3>\n
          First, we\u2019ll use Github Codespace and Python in conjunction with the OpenAI GPT-3.5 Turbo<\/strong> model for natural language processing. Make sure you have the required Python packages installed and obtain your OpenAI API and Comet API keys. (Reference: OpenAI Help Center \u2014 Where can I find my API key?<\/em><\/a>; CometLLM \u2014 Obtaining your API key<\/em><\/a>)<\/p>\n
          <\/div>\n
          The provided code snippet initializes the required components.<\/p>\n
          import<\/span> os\nfrom<\/span> openai import<\/span> OpenAI<\/span>\nimport<\/span> comet_llm\nfrom<\/span> IPython<\/span>.display<\/span> import<\/span> display\nimport<\/span> ipywidgets as<\/span> widgets\nimport<\/span> time\nimport<\/span> dotenv <\/span><\/pre>\n
          # Input API keys<\/span>\n\nMY_OPENAI_KEY = os.getenv('MY_OPENAI_KEY'<\/span>)\nMY_COMET_KEY = os.getenv('MY_COMET_KEY'<\/span>)\n\n\ncomet_llm.init(project=\"ecom_shopbot_openai\"<\/span>,\n               api_key=MY_COMET_KEY)<\/span><\/pre>\n
          3. Defining Product Information<\/h3>\n
          The e-commerce chatbot needs information about the products available in your store. Product list includes details such as name, price, available sizes, colors, etc. This information is stored in a structured format, making it easier for chatbots to provide accurate details to customers.<\/p>\n
          product_list = '''\n# Fashion Shop Product List\n\n## Women's Clothing:\n- T-shirt\n  - Price: $20\n  - Available Sizes: Small, Medium, Large, XL\n  - Available Colors: Red, White, Black, Gray, Navy\n\n# ... (Other product categories and details)\n'''<\/span><\/span><\/pre>\n
          <\/div>\n
          This step allows you to enter the product list for your store. Keep in mind that the more complex your product list, the more detailed its context<\/code> needs to be so that the generative AI can clearly understand its requirements.<\/p>\n
          context = [{'role'<\/span>: 'system'<\/span>,\n            'content'<\/span>: f\"\"\"\nYou are ShopBot, an AI assistant for my online fashion shop - Nhi Yen.\n\nYour role is to assist customers in browsing products, providing information, and guiding them through the checkout process.\n\nBe friendly and helpful in your interactions.\n\nWe offer a variety of products across categories such as Women's Clothing, Men's clothing, Accessories, Kids' Collection, Footwears and Activewear products.\n\nFeel free to ask customers about their preferences, recommend products, and inform them about any ongoing promotions.\n\nThe Current Product List is limited as below:\n\n```{product_list}<\/span>```\n\nMake the shopping experience enjoyable and encourage customers to reach out if they have any questions or need assistance.\n\"\"\"<\/span>}]<\/span><\/pre>\n
          4. Creating the Chatbot<\/h3>\n
          After you configure your environment and define your product information, you can create your ShopBot chatbot. The get_completion_from_messages<\/code> function sends messages to the OpenAI GPT-3.5 Turbo model and retrieves responses from the model.<\/p>\n
          def get_completion_from_messages<\/span>(messages, model=\"gpt-3.5-turbo\"<\/span>)<\/span>:\n    client =<\/span> OpenAI<\/span>(\n        api_key=MY_OPENAI_KEY,\n    )\n\n    chat_completion = client.chat.completions.create<\/span>(\n        messages=messages,\n        model=model,\n    )\n    return<\/span> chat_completion.choices[0<\/span>].message.content<\/span><\/pre>\n
          5. Interacting with Customers<\/h3>\n
          To interact with customers, use a simple user interface with text entry for customer messages and a button to start a conversation. The collect_messages<\/code> function processes user input, updates conversation context, and displays chat history.<\/p>\n
          # Interacting with Customers<\/span>\n\ndef<\/span> collect_messages<\/span>(_<\/span>):\n    user_input = inp.value\n    inp.value = ''<\/span>\n    context.append({'role'<\/span>:'user'<\/span>, 'content'<\/span>:f\"{user_input}<\/span>\"<\/span>})\n\n    # Record the start time<\/span>\n    start_time = time.time()\n\n    response = get_completion_from_messages(context)\n\n    # Record the end time<\/span>\n    end_time = time.time()\n\n    # Calculate the duration<\/span>\n    duration = end_time - start_time\n\n    # Log to comet_llm<\/span>\n    comet_llm.log_prompt(\n        prompt=user_input,\n        output=response,\n        duration=duration,\n        metadata={\n        \"role\"<\/span>: context[-1<\/span>]['role'<\/span>],\n        \"content\"<\/span>: context[-1<\/span>]['content'<\/span>],\n        \"context\"<\/span>: context,\n        \"product_list\"<\/span>: product_list\n        },\n\n    )\n\n    context.append({'role'<\/span>: 'assistant'<\/span>, 'content'<\/span>: f\"{response}<\/span>\"<\/span>})\n\n    user_pane = widgets.Output()\n    with<\/span> user_pane:\n        display(widgets.HTML(f\"<b>User:<\/b> {user_input}<\/span>\"<\/span>))\n\n    assistant_pane = widgets.Output()\n    with<\/span> assistant_pane:\n        display(widgets.HTML(f\"<b>Assistant:<\/b> {response}<\/span>\"<\/span>))\n\n    display(widgets.VBox([user_pane, assistant_pane]))\n\ninp = widgets.Text(value=\"Hi\"<\/span>, placeholder='Enter text here\u2026'<\/span>)\nbutton_conversation = widgets.Button(description=\"Chat!\"<\/span>)\nbutton_conversation.on_click(collect_messages)\n\ndashboard = widgets.VBox([inp, button_conversation])\n\ndisplay(dashboard)<\/span><\/pre>\n
          Let\u2019s play with the bot. Tailor the context to your own needs by following the Prompting for Developers: A Guideline<\/strong> above.<\/p>\n
          \n
          \n
          \"\"<\/figure>
          <\/picture><\/div>\n<\/div>
          Demo Shopbot (Image by Author)<\/figcaption><\/figure>\n
          In this project, I\u2019m using comet_llm<\/code> to keep track of what users ask, how the bot responds, and how long each interaction takes. The information is logged to this comet project<\/em><\/strong><\/a>. This helps in improving the model for future training. You can learn more about experiment tracking with Comet LLM<\/a>.<\/p>\n
              # Log to comet_llm<\/span>\n    comet_llm.log_prompt(\n        prompt=user_input,\n        output=response,\n        duration=duration,\n        metadata={\n        \"role\"<\/span>: context[-1<\/span>]['role'<\/span>],\n        \"content\"<\/span>: context[-1<\/span>]['content'<\/span>],\n        \"context\"<\/span>: context,\n        \"product_list\"<\/span>: product_list\n        },\n    )<\/span><\/pre>\n
          The logs in the Comet UI appear like this. By analyzing these logs, you can use various strategies to make responses quicker, improve accuracy, enhance customer satisfaction, and eliminate unnecessary steps in your ecommerce operations.<\/p>\n
          \n
          \n
          \"\"<\/figure>
          <\/picture><\/div>\n<\/div>\n
          Further LLM training based on the logs (Image by Author)<\/figcaption>\n<\/figure>\n
          Additional<\/em><\/strong>: Summarize the order details and send to the user and the respective departments as follows:<\/p>\n
          messages =  context.copy<\/span>()\nmessages.append<\/span>(\n{'role'<\/span>:'system'<\/span>\n , 'content'<\/span>:'create a table summary of the previous order.'<\/span>}\n)\n\nresponse = get_completion_from_messages(messages)\nprint<\/span>(response)<\/span><\/pre>\n
          Output:<\/p>\n
          Certainly!<\/span> Here's<\/span> a summary of your previous order:<\/span>\n\n|<\/span>     Product<\/span> Name<\/span>    |<\/span>   Type<\/span>   |<\/span>   Size<\/span>   |<\/span>   Color<\/span>   |<\/span>   Price<\/span>   |<\/span>\n|---------------------|----------|----------|-----------|-----------|<\/span>\n|<\/span>     Red<\/span> T-shirt<\/span>     |<\/span>  Women<\/span>   |<\/span>   Small<\/span>  |<\/span>    Red<\/span>    |<\/span>   $20<\/span>     |<\/span>\n|<\/span> Comfortable<\/span> Sneakers|<\/span>  Women<\/span>   |<\/span>   12<\/span>     |<\/span>   Black<\/span>   |<\/span>   $55<\/span>     |<\/span>\n\nIf<\/span> you<\/span> have<\/span> any<\/span> other<\/span> questions<\/span> or<\/span> need<\/span> further<\/span> assistance,<\/span> please<\/span> let<\/span> me<\/span> know.<\/span><\/span><\/pre>\n
          III. Reference<\/strong><\/h2>\n