SEI this week announced the launch of SEIGPT, its new framework for building generative AI applications.
For now, the technology, investment and asset management provider for advisors is only using new tools and applications powered by SEIGPT to increase efficiency and reduce cost in areas that include customer service, process automation, human resources, analytics of contracts and code development. It is an outgrowth of SEI's AI Center of Excellence, launched in 2023 and led by AI Director Vlad Coric.
“We have approximately 400 employees using SEIGPT applications in production and pilot across various business areas, including customer service, middle office, marketing, human resources, legal, operations and development,” Coric wrote in an email.
SEIGPT is based on a feedback-augmented generation architecture, which provides precise references to the data used to create a response and thus provides greater transparency, Coric said. As a result, the RAG framework offers an attractive architecture for companies.
While the new technology is developed in-house, SEI used some Microsoft components, such as their OpenAI service and AI research service, according to Coric.
As the name might suggest, SEI is using OpenAI's Generic Pre-Trained Transformers (GPT) to apply the power of large linguistic models to its massive amounts of data, which are spread across various data sources.
By using the new framework and applying advanced natural language processing capabilities to all that data, the firm aims to improve its search functionality, support content creation (including personalized content), streamline workflows , answer user questions and provide automated summaries of information and data.
According to SEI's chief data officer Deepak Bhardwaj, the new AI-powered apps — seven of which are currently in production — have helped the firm's employees respond to more than 7,000 customer service requests and analyze dozens of thousands of pages of legal contracts so far. Five more SEIGPT-powered applications are on the pipeline for domestic submission by the end of 2024.
“The benefit of optimization extends beyond SEI to our customers,” SEI Chief Technology Officer Zach Womack said in a statement announcing the technology.
“The sooner we can provide information or address a concern, the more time our employees and customers have to focus on what they do best: customer service,” he said.
In addition to launching the AI Center of Excellence last year, SEI invested $10 million in AI developer and incubator TIFIN in February.
SEI manages, advises or administers approximately $1.6 trillion in assets.