SOOP is using artificial intelligence (AI) to enhance the live streaming experience. It is applying AI technology throughout the streaming process—such as broadcast operations support, content discovery, and user-participatory video production—to broaden touchpoints between streamers and users.
On the 2nd, industry sources said SOOP is expanding AI features across its live streaming service. A representative example is the AI manager "SARSA (SOOP AI Realtime Streaming Assistant)," which helps streamers run broadcasts more smoothly and communicate with users.
SARSA provides a variety of features—including broadcast setup support, AI broadcast summaries, chat management, chat atmosphere analysis, and feature recommendations—to assist streamers with broadcast operations. It helps with operational tasks and analyzes broadcast situations so streamers can focus on content creation and communicating with users.
In the recently released SARSA 2.0, a function was added that lets the AI communicate directly with users and keep the broadcast flow going. When a streamer steps away briefly or runs a sleeping stream, the AI chats with users, recommends content, or watches video-on-demand (VOD) together to maintain the flow of the broadcast.
In particular, SARSA is implemented as a persona-based AI manager that learns a streamer's VOD and broadcast data to reflect the speaking style and broadcast atmosphere. SOOP expects this will allow the communication flow within a broadcast to be maintained even when the streamer is absent.
SOOP is also running "SOOPi (Soop Personalized Intelligence)," an AI video assistant that helps users discover content. SOOPi is an AI trained on a streamer's face, voice, speaking style, and broadcast history, and it recommends content and provides related information while conversing with users in real time.
For example, users can ask about a specific streamer's broadcast style, key content, or past broadcasts, and the AI recommends related videos or provides needed information. In addition, with the recent addition of long-term memory and performance improvements based on a large video model (LVM), it can remember transfer conversations and enable more natural, context-aware interactions.
It currently supports a total of 54 streamers, including Woowakgood, Lee Sang-ho, and Kim Min-gyo, and plans to expand the scope of support going forward.
SOOP is also expanding an environment where users can directly participate in content through AI. The Generative AI video maker "SAVYG (Soop AI Video Yielding Generative)" is a service that converts text and photos into video, based on "MotionGPT," an AI technology trained on streamers' video data.
Users can create AI content with SAVYG, such as reaction videos that use a streamer's facial expressions and gestures. Its flagship feature, "SAVYG video balloon," lets users gift the videos they create to streamers or enjoy them together during a broadcast.
SOOP plans to broaden participation methods within the live streaming ecosystem by expanding AI use into broadcast operations support, content discovery, and participation in content creation.