Perplexity and NVIDIA are bringing “Portable Computer,” an autonomous AI spy solution that prioritizes data security, to the Windows ecosystem. Previously only available on Linux, the system leverages the local processing power of NVIDIA GeForce RTX and RTX PRO series graphics cards to process sensitive documents, codebases, and financial reports directly on the user’s hardware without uploading them to the cloud.

Local Power, Complete Data Security: What is Portable Computer?

While the analytical and automation capabilities of AI models accelerate daily workflows, transferring corporate and personal data to cloud servers raises significant privacy concerns. Perplexity’s Portable Computer, developed in partnership with NVIDIA, addresses this issue by delegating complex and multi-step tasks to an AI spy running on the user’s computer.

Because the system runs on local hardware, it does not consume any random API or cloud credits. While documents remain within the computer’s internal storage during the process, the user has absolute control over which data goes externally. The AI ​​spy requests explicit permission from the user before establishing external contact if it needs to perform a web search or if more complex cloud-based reasoning that exceeds local capacity is required.

Featured Use Cases

The Portable Computer can plan and complete multi-step workflows like a direct colleague in a desktop environment. The main scenarios undertaken by the spy on local hardware include:

• Code and Repository Reviews: Connects to developers’ GitHub repositories, analyzes open pull requests, identifies missing or outdated documentation, and lists the next steps.
• Precise Financial Analysis: Scans closed documents locally, such as bank statements, tax records, and company balance sheets. Reports duplicate or incorrect entries with reference to the source document and page number.
• Workflows and Insight Extraction: Processes complex datasets on the device, such as customer churn analyzes or record steps, and directly transmits critical findings to the group’s Slack channels.

Hardware Requirements and System Compatibility:

Running advanced AI models smoothly on the device requires significant memory capacity. Accordingly, Windows support is offered for NVIDIA GeForce RTX (eg, RTX 3090, RTX 4090 series) and enterprise RTX PRO class graphics processors with at least 24 GB of VRAM capacity. Tensor cores provide the inference speed required by the spy architecture, enabling large language models to run smoothly on local hardware.

A New Era in the Native AI Ecosystem

This expansion to the Windows ecosystem after the Linux environment represents a valuable milestone in integrating autonomous AI tools into daily professional workflows. The convergence of NVIDIA’s hardware leadership in the desktop AI market with Perplexity’s software infrastructure; Eliminating cloud costs while demonstrating that high-performance local work models based on data privacy will become the standard in the future.