Consumer Reports (CR), renowned for its commitment to providing unbiased product information, has launched AskCR, a groundbreaking conversational AI tool designed to assist consumers with personalized product recommendations. This initiative, developed in collaboration with NineTwoThree, represents a significant advancement in the application of generative AI for consumer research.
A Cutting-Edge Collaboration
NineTwoThree played a crucial role in the development of AskCR, working closely with CR's Innovation Lab to create a sophisticated AI-driven chatbot. This experimental tool leverages CR’s extensive database of product ratings, reviews, and recommendations to deliver tailored advice and insights to users.
Project Highlights
- Design & Development
NineTwoThree's expertise in AI/ML engineering and UI/UX design was pivotal in crafting AskCR’s user experience. The team designed a conversational interface capable of handling complex queries and providing actionable product recommendations. - RAG System Architecture
The core of AskCR is its custom-built Retrieval-Augmented Generation (RAG) system. Unlike off-the-shelf solutions, NineTwoThree tailored this system to address CR’s unique requirements, enhancing its ability to retrieve accurate information from a vast array of data sources. - Collaborative Engineering
The project involved seamless collaboration between NineTwoThree and CR’s engineering and product teams. Regular meetings, peer code reviews, and strategic sessions ensured the project’s success. The teams employed GitHub for version control, Jira for project management, and Slack for ongoing communication. - Security and Evaluation
NineTwoThree implemented robust security measures and evaluation processes to ensure the reliability and safety of AskCR. This included rigorous testing for both functional performance and security vulnerabilities.
Innovative Solutions and Future Outlook
The development of AskCR required innovative approaches to address several challenges:
- Customized AI Solutions
Standard RAG frameworks were insufficient for CR’s needs. NineTwoThree customized their approach to achieve higher accuracy in answering user queries, leveraging detailed instructions and dynamic data pipelines. - Enhanced User Interaction
The team developed a sophisticated query refinement and routing system to interpret and respond to user questions effectively. This involved balancing simplicity with the need for high-quality, accurate responses. - Memory Integration
To improve the AI’s contextual understanding, NineTwoThree implemented memory features that allow AskCR to retain and utilize information from past interactions. - User Testing and Load Management
Extensive user testing and load simulations were conducted to ensure AskCR’s performance under various conditions, providing valuable feedback for continuous improvement.
