#IndexConf 2023 - How We Built AI Powered Workplace Search at Glean

Rockset
Rockset
4 هزار بار بازدید - 10 ماه پیش - Speakers: T.R. Vishwanath, Co-Founder; James
Speakers: T.R. Vishwanath, Co-Founder; James Simonsen, Software Engineer Glean introduced enterprise search using AI, leveraging deep learning techniques for enterprise knowledge understanding and personalization to provide relevant answers to user questions. In this talk, T.R. Vishwanath and James Simonsen discuss how they built an enterprise knowledge graph that indexes content, people and activity to serve fast search results. They’ll talk through the challenges they faced making AI-based search real time and privacy aware at scale. Learn more about the Index Conference: rockset.com/index-conf/ Here's the chapterization of the video with timestamps for YouTube: 0:00 - Introduction to the session and the power of AI in workplace search. 0:39 - Welcome and introduction of speakers TR Vishwanath and James Simonson from Glean. 1:22 - Overview of the challenges in workplace search. 3:49 - Initial thoughts on building the search solution using existing APIs. 4:42 - Challenges with federated search and moving towards a unified indexing approach. 5:34 - Addressing security and permission issues when implementing workplace search. 7:20 - Tackling the issue of stale data and ensuring the freshness of the search index. 9:12 - Scaling the search solution for large organizations. 10:34 - Technical decisions on scaling different components of the system. 11:27 - Managing API quota and optimizing crawl strategies. 12:27 - Ensuring security and compliance with regulations. 13:40 - Improving search relevance and dealing with enterprise-specific challenges. 16:27 - Customizing search algorithms to adapt to unique enterprise terminologies. 18:11 - Extending capabilities with AI-powered assistants and query answering. 19:44 - Discussion on leveraging workplace search for broader applications. 21:26 - Final Q&A with audience, discussing deployment models and the effectiveness of Glean. 24:19 - Further questions on API access and deployment flexibility. 25:36 - Addressing specific use cases and advantages over native search functionalities. 27:26 - Answering online questions about multilingual support and technical details. 29:14 - Closing remarks and thanks to the audience.
10 ماه پیش در تاریخ 1402/08/16 منتشر شده است.
4,000 بـار بازدید شده
... بیشتر