Workflow
Amazon internal review found its AI tool Q fell 'significantly' behind rivals on accuracy in its first year
AmazonAmazon(US:AMZN) Business Insiderยท2025-10-03 09:00

Core Insights - Amazon's AI productivity tool, Q Business, has faced significant challenges in accuracy and performance, falling behind competitors in key features [1][4][12] - The internal document highlights customer complaints regarding the tool's inability to process non-text data and maintain conversational flow [2][11][12] - Despite these issues, Amazon has implemented a formal accuracy program and rolled out updates to improve the tool's performance [14][15] Group 1: Performance Challenges - Q Business struggled with processing tabular and spreadsheet data, leading to customer dissatisfaction from companies like Accenture and Intuit [1][11] - The tool's conversational abilities lagged behind competitors, resulting in incomplete responses and difficulties in maintaining context during interactions [12][19] - Staffing issues within the accuracy team contributed to the challenges faced by Q Business, with multiple product manager changes and inadequate resources reported [13] Group 2: Improvement Efforts - Amazon initiated a formal accuracy program in February to address the challenges faced by Q Business [14] - Updates rolled out included a "hallucination mitigation" feature and a response customization tool, aimed at enhancing the chat experience and query answering capabilities [14][15] - Positive feedback has been received from several customers, including Nasdaq and Jabil, indicating some success in improving the tool's utility [16][17][18] Group 3: Market Position and Future Outlook - Q Business achieved 90% accuracy for text-rich data, but doubts about its future persist among some employees, citing a lack of strong performance in business applications [19] - Amazon's other AI offerings have also struggled in terms of revenue compared to competitors, prompting a reevaluation of sales strategies [20] - An AWS spokesperson defended the company's position, claiming success in various AI applications beyond infrastructure [20][21]