Palantir fed Nvidia customer meeting transcripts into ontology memory to retrain models

pdamodaran · x · 2026-09-14

The most striking takeaway from this year's aipcon: Palantir fed customer meeting transcripts (with the customer's, i.e. Nvidia's, permission) into its ontology memory to retrain models. The author argues this has massive implications for the knowledge-work space — real enterprise conversations becoming a training asset.

Original post →

More from Companies & People

Companies & People channel →