AI-Powered Sales Advisor for Ecommerce

Project background

Our R&D department developed this experimental solution to explore how AI-driven personalization can solve challenges in retail. We created a functional AI assistant that acts as a virtual sales advisor, offering the user tailored product recommendations.

The solution uses hardcoded data to provide targeted advice within specific product categories.

We tested it successfully in two scenarios: a car rental advisor that helped users choose the best rental options, and a multicooker advisor that recommended kitchen appliances across three price ranges: premium, mid-range, and budget.

Customer challenges

Modern ecommerce solutions aim to be as easy and comfortable to use as possible. However, online customers still face a number of challenges:

• Information overload: With the overwhelming number of available products, customers can find it difficult to make decisions.
• Customer service efficiency: Customers often experience delays and inconsistent responses when seeking support, as retailers struggle to handle multiple inquiries at the same time
• Unsatisfactory personalization: Modern consumers expect a high degree of personalization in their shopping journey and they will choose the platform that understands them best.

  

Business challenges

As commerce goes online, retailers globally face increasing competition. Their top struggles include:

• Customer engagement: Retailers struggle to keep customers engaged due to a lack of personalized experiences.
• Data utilization: Although retailers collect a large amount of data, they often fail to leverage it effectively for personalized insights.
• Omnichannel retail integration: Operating seamlessly across multiple retail channels (online, in-store, mobile) is a challenge.
• Product promotion flexibility: Retailers need the flexibility to promote specific products as part of marketing or sales strategies.

  

Solution

Our newly-developed assistant uses advanced AI to analyze customer preferences, process product data, and deliver personalized recommendations.

Key components of the solution include:

• AI-powered chatbot: Integrated with a private instance of Azure OpenAI model called GPT-4, the chatbot handles customer inquiries, offering personalized product suggestions and acting as a virtual sales advisor. It uses a vector database to retrieve relevant information and generate intelligent responses.
• Personalized recommendations: The assistant suggests products in top, medium, and budget price tiers by analyzing product data stored in a vector database for quick access.
• Multilingual and omnichannel support: The assistant works across platforms (web, mobile, in-store) and supports multiple languages, making it accessible to a global audience.

Technological stack

• Frontend: Built with React.js for flexibility and developer support.
• Backend: Powered by Python, with an admin panel for managing recommendations.
• Cloud infrastructure: Hosted on AWS, ensuring scalability and reliability.
• Azure OpenAI integration: GPT-4 provides natural language responses for a smooth user experience.

  

Results

The AI assistant helps retailers enhance customer experience, increase satisfaction, and conversions.

The solution offers:

• Omnichannel experience: It ensures a consistent, personalized experience for customers across all retail channels.
• Data-driven insights: The system processes both hardcoded and live product data, delivering real-time recommendations aligned with customer preferences and trends.
• Promotion mechanism: Administrators can promote specific products through an integrated recommendation system, allowing for targeted marketing efforts based on sales priorities.