Interview
summary
Salesforce teams deploy Email Insights to spot real‑time signals (pricing talks, competitor mentions, even weather) from hundreds of TB and billions of records, and build a contacts graph from communication logs to power call‑recommend and cross‑company connection features. Pipeline uses entity resolution, distributed graph processing, NLP, deep learning and functional programming patterns that ship functions to data. Notes present concrete production challenges: tooling gaps moving research to Einstein production, surprising scale costs, and evaluation numbers for latency and precision on large sparse graphs.
AI-written summary · about
related talks
07mThomas Gerber Interview
35mUsing AI to Fuel Digital Transformation & Growth
29mIntelligent Log Analysis and Real-time Anomaly Detection @ Salesforce
46mIntroducing Salesforce Genie: A real-time platform for customer magic
33mUnleash the Magic of Salesforce Genie - in Action
30mArtificial Intelligence More Like Artificial Stupidity
47mMetrics Driven Machine Learning Development at Salesforce Einstein
03mInterview