Perplexity's Personal Computer splits AI tasks between device and cloud
New hybrid system keeps private data local while using cloud for heavy tasks.
Perplexity announced a new hybrid local-server system for its Personal Computer AI agent that can split tasks between a local model on your device and frontier models in the cloud. Starting in July, the system automatically decides which parts of a task run locally—keeping sensitive data like financial records, health information, and personal files on your device—and which get sent to more powerful cloud models. Built with Intel and also supporting Nvidia’s RTX Spark platform, it maximizes token efficiency and cuts cloud computing costs. Available now on Mac, with Windows support coming.
- Automatic task routing between local model (sensitive data) and cloud model (complex reasoning)
- Supports Intel and Nvidia RTX Spark hardware for on-device inference
- Available on Mac starting July 2026, with Windows version confirmed
Why It Matters
Enables privacy-preserving AI agents that reduce cloud costs while maintaining access to top-tier models.