Perplexity announces hybrid inference system that routes AI tasks between local devices and cloud models, starting in July for its Personal Computer agent

Perplexity introduced an orchestrator that automatically decides whether to process tasks on a user's device or in the cloud, optimizing for accuracy, privacy, and energy efficiency. The system will be integrated into the Always-on agent product Personal Computer, which was introduced in March, starting in July.

Sensitive data such as financial documents or health information stays on the local device, while compute-intensive tasks are routed to cloud models. The model-agnostic framework runs on Intel hardware as well as other hardware like Nvidia's RTX Spark.

According to Perplexity, shifting routine tasks to local devices could reduce the need for centralized computing infrastructure and simplify data sovereignty questions. The company says its business model rewards correct answers instead of high compute consumption, making efficiency optimization a natural incentive.