PLTR · Earnings call · 2026Q3T · Full report

AI Platform Positioning

Palantir Technologies Inc. · 2026-11-02 · Importance null · Surprise 24

Shyam Sankar positioned AIP as an enterprise workflow platform that integrates AI, software and people to generate measurable operational value, rather than simply automating individual knowledge workers. He said post-training models on customers’ operational data in the Ontology can produce task-specific performance beyond frontier models while keeping customer-controlled weights and data. AutoResearch and AIP Evolve are intended to automate model building, deployment and evaluation, while the post-training pipeline compounds value through repeated operational-data feedback.

Key facts

• Palantir and Tampa General improved model performance by 7% and reduced latency by 22x using fine-tuned and post-trained Nemotron.
• Palantir's AIP and AIG implemented AIG's automation of insurance underwriting for complex commercial lines.
• Palantir's CTO said a customer at DevCon built a production-ready model that was better than frontier for its use case in 48 hours as a hack-week project on its own stack.
• Palantir's AutoResearch feature and continued investment in AIP Evolve are intended to move toward an automated loop that continuously builds, deploys, and evaluates intelligence against ROI.
• Palantir said McCarthy Building Companies will save 30,000 hours annually using post-trained Nemotron to help turn project requirements into contract language.
• Palantir's AIP implements enterprise workflows and integrates AI, software, and people to generate objective, quantifiable value.