In a bold move towards innovation and efficiency, Amazon has launched a revolutionary sales assistant that promises to optimize the advertising process on Twitch, the popular live streaming platform. This project, launched in early 2024, represents a significant effort by Amazon to maximize Twitch’s potential as a powerful advertising channel.
Twitch, which attracts over 105 million monthly visitors, has faced significant challenges in providing real-time advertising information. Sales teams encountered a complex array of advertising products that often hindered quick and efficient access to the necessary information to close deals. Questions asked in these public channels took an average of two hours to be answered, if they were answered at all.
To address these obstacles, Amazon has developed an application called Retrieval Augmented Generation (RAG) implemented in a Slack-based chat assistant. The solution leverages Amazon Bedrock and advanced language models, integrating with Twitch’s vast internal marketing documentation. This system allows sellers to instantly access relevant information, reducing response time and facilitating the launch of campaigns in a minimal time frame.
The heart of this technology is a vector database that organizes all relevant information, from wiki documents to PowerPoint presentations, periodically updating to ensure that the data is always up to date and ready to be used by the machine learning model.
The use of the RAG application has not only reduced the time for implementing advertising campaigns but has also increased the effectiveness of the sales team. Since its launch in February, the chatbot has handled over 11,000 queries about the sales process on Twitch. As a result of this tool, Twitch sales managed by this system have increased by 25% compared to the previous year.
The key to this success lies in its innovative agent workflow. This system detects the nature of each query, from audience forecasts to specific advertising products, and decides the best source of information to use. Thus, the assistant not only responds but also evaluates and selects the appropriate tools to provide accurate and contextual answers, achieving dynamic integrations between Amazon and Twitch’s advertising products.
Amazon continues its path of innovation, seeking to expand this system to other advertising services of the company, with plans to incorporate knowledge bases of other key products such as Prime Video and Alexa. This expansion not only promises to make internal operations smoother and faster but also redefines the way advertising sales on impactful digital platforms are perceived.
The implementation of this solution at Amazon not only enhances commercial efficiency but also encourages other companies to consider similar strategies to address corporate challenges through artificial intelligence and machine learning, setting a new standard in technology and digital sales.
via: MiMub in Spanish
