Overview
Syllabus
0:00 Intro
01:00 What Glean is and how it works
02:39 Starting Glean before the LLM boom
04:10 Using transformers early in enterprise search
06:48 Semantic search vs. generative answers
08:13 When to fine-tune vs. use out-of-box models
12:38 The value of small, purpose-trained models
13:04 Enterprise security and embedding risks
16:31 Lessons from Rubrik and starting Glean
19:31 The contrarian bet on enterprise search
22:57 Culture and lessons learned from Google
25:13 Everyone will have their own AI-powered "team"
28:43 Using AI to keep documentation evergreen
31:22 AI-generated churn and risk analysis
33:55 Measuring model improvement with golden sets
36:05 Suppressing hallucinations with citations
39:22 Agents that can ping humans for help
40:41 AI as a force multiplier, not a replacement
42:26 The enduring value of hard work
Taught by
Weights & Biases