The video introduces Perplexity’s Hybrid Compute feature, which seamlessly combines local AI processing on Apple Silicon Macs with cloud AI services to balance powerful performance and strict data privacy through a built-in privacy guard. It demonstrates how this hybrid approach enables secure, efficient AI tasks—such as analyzing sensitive client data—by keeping private information local while leveraging up-to-date cloud models, all within an easy-to-use app.
The video discusses the division of AI usage into two main categories: online AI, which operates in large data centers and offers powerful, fast processing but is remote, and local AI, which allows users to maintain full control over their data and AI models on their own devices. The presenter introduces the concept of a hybrid AI setup that combines the strengths of both approaches—keeping sensitive data local while leveraging cloud resources for more demanding tasks. Although such a setup might seem complex, the video highlights that Perplexity’s Hybrid Compute feature simplifies this process significantly.
Perplexity’s new Mac OS app includes the Hybrid Compute feature, which seamlessly integrates cloud AI services with local AI models while ensuring privacy through a built-in privacy guard. This guard uses an on-device classifier to detect sensitive information and prevent it from being sent to the cloud unless explicitly permitted by the user. The app supports Apple Silicon Macs with at least 24GB of RAM and offers an easy installation and setup process, including downloading appropriate AI models based on the user’s hardware capabilities.
Once installed, users can switch between different modes within the app, such as search and compute, and configure the orchestrator to use local, cloud, or hybrid compute. The privacy guard is a standout feature, allowing users to mask sensitive data, keep processing local, or request permission before sending data to the cloud. The models used are open source and available on Hugging Face, ensuring transparency and flexibility for users.
The video demonstrates a practical use case where an IT consultant uses Perplexity Hybrid Compute to analyze a client list containing sensitive information and generate personalized Apple Mac upgrade recommendations. The consultant’s private data remains securely on the Mac, while the AI accesses up-to-date cloud information to provide accurate and detailed advice. The app also provides real-time system resource monitoring during local processing, enhancing user confidence in the tool’s efficiency and security.
In conclusion, Perplexity Hybrid Compute offers a powerful and user-friendly solution that leverages both local AI capabilities and cloud resources while maintaining strict data privacy. It enables users to make full use of their Apple Silicon hardware and access frontier AI models safely. The video encourages viewers to try the app via a provided link, emphasizing the ease of setup and the benefits of combining local and cloud AI for enhanced productivity and privacy.