QuickAssist AI

Inspiration

Driven by the vision to transform online commerce communication, our project seeks to address a critical void in the messaging platforms of major e-commerce hubs. Starting with eBay, which supports a vast network of over 18 million sellers, we identified a pressing need for more sophisticated tools that could drastically improve the efficiency and effectiveness of buyer-seller interactions.

Moving beyond eBay, we turned our attention to WhatsApp Business, a platform boasting over 200 million monthly active users. Despite its widespread use among small businesses globally, the platform's communication tools are surprisingly limited. This realization prompted us to consider how AI functionalities could be integrated to enhance the seller experience on WhatsApp Web.

Similarly, on Shopify Inbox, which serves over 900,000 Shopify stores, there's a clear appetite for innovative solutions that could streamline communication between sellers and buyers. Addressing common communication challenges, such as providing quick responses and refining message content, can significantly empower sellers to provide a smoother support experience.

Our comprehensive analysis across various platforms, including eBay, WhatsApp Web, and Shopify Inbox, pinpointed a universal challenge: the absence of AI-driven tools to aid sellers in crafting quick, professional, and effortless responses. The time lost by small businesses on repetitive tasks, such as message drafting, grammar checking, and tone maintenance, is substantial.

This gap in the market led us to create QuickAssist AI, a Chrome extension designed to integrate flawlessly with these platforms. Leveraging Google Chrome’s Gemini Nano, QuickAssist AI provides real-time, on-the-spot assistance to enhance seller communication within their existing workflows. By performing all processing directly on the client side, it ensures robust data privacy, as sensitive information never leaves the user's device. This unique approach enables seamless, efficient, and secure communication for sellers.

What it does

QuickAssist AI serves as a tool that elevates the level of communication between sellers and buyers on platforms such as eBay, Shopify Inbox and WhatsApp Web. It provides a suite of AI-enhanced features designed to optimize customer interactions and streamline the communication process. Here's a closer look at how QuickAssist AI seller-customer exchanges:

Quick Replies: Generate instant, context-aware professional responses. The AI analyzes the context of the conversation and suggests appropriate responses that maintain a professional tone, ensuring that customer relations are handled with the utmost efficiency and care.
Improve Text: Sophisticated response refinement tool that works to improve the grammar and clarity of seller responses. Whether it's polishing an existing reply or ensuring that your first draft is free from errors.
Voice Input: Leveraging the capabilities of the Web Speech API, this feature allows users to dictate their messages, which QuickAssist AI then converts into text. This is particularly useful for sellers who are on the go or find typing cumbersome (Currently the functionality is implemented on eBay & Shopify Inbox)
Canned Responses: For frequent and repetitive questions, QuickAssist AI provides the ability to save and use canned responses. This means that sellers can create a repository of well-crafted replies to common queries, which can be reused with just a click.

How we built it

QuickAssist AI is built as a Chrome extension and uses vanilla Javascript and Chrome APIs. The key components include:

Frontend: Vanilla JavaScript and CSS were used to create a clean and intuitive interface.
Integration: Integrated directly with platform-specific elements. For instance, on WhatsApp, it hooks into chat input areas, on Shopify inbox and on eBay, it integrates with the message box interface. This is achieved through the use of content scripts, which are a crucial part of the Chrome extension framework, allowing the extension to interact with the content of web pages dynamically.
Functionality: At the core of QuickAssist AI is its integration of Gemini Nano AI models, delivering context-aware suggestions and corrections for features like Quick Replies and Text Improvement. This AI-driven functionality is further enhanced by the Web Speech API, enabling seamless voice-to-text conversion to make response drafting faster and more intuitive. Supporting these features, IndexedDB plays a vital role by efficiently managing canned responses, allowing sellers to store and quickly access pre-written replies within their workflows.
Customization: Added platform-specific buttons and icons for generating responses & response improvement. These elements were carefully designed to match the aesthetic and theme of the respective platforms, ensuring that QuickAssist AI doesn't feel out of place.

Challenges we ran into

Tailoring the extension to work seamlessly across diverse platforms like WhatsApp Web, eBay, and Shopify Inbox presented unique challenges. Each platform's UI required customized handling for text. For instance, WhatsApp Web has state management in place, which disables direct text manipulation and demands advanced DOM traversal techniques to interact and update effectively with its components. Adapting to these differences required a deep understanding of platform-specific architectures.

Accomplishments that we're proud of

We are proud of the extension's intuitive design, which allows sellers to adopt it effortlessly into their daily workflows across various platforms. It's optimized for speed, ensuring that the AI-powered assistance it offers doesn't hinder but rather accelerates response times. The real accomplishment, however, is its practical utility: QuickAssist AI is an indispensable tool for customer support professionals, enabling them to provide faster, more accurate assistance, thereby improving the overall customer experience.

What we learned

Building QuickAssist AI was a fun and insightful journey. We gained hands-on experience with Chrome's built-in AI models, explored the powerful capabilities of the Web Speech API for voice input, and delved into various database options available in Chrome for data storage and retrieval.

What's next for QuickAssist AI

Client Side RAG / Fine-tuning: Explore approaches for integrating RAG( (Retrieval-Augmented Generation) on the client side using indexedDB or options that let the Gemini nano model be fine-tuned for different businesses on their domain and products.
Platform Expansion: Extend compatibility to additional platforms used by businesses for customer support.
Advanced AI Features: To enhance the extension's helpfulness, we intend to incorporate advanced AI functionalities, such as offering multiple contextually relevant responses or auto suggestions. This will provide users with a range of options to choose from, ensuring they can select the most suitable reply to address customer queries effectively and with the right tone.
Multi-Language Support: Integrate multilingual capabilities to cater to a global audience and facilitate communication in various languages by exploring Translation API.