.png){kind=logo}
Local-first AI: Users can run models and AI agents directly on compatible NVIDIA hardware without constantly sending data to the cloud.
Hybrid intelligence: Portable Computer can combine private local processing with stronger cloud models when users approve external data sharing.
Agent-focused computing: The platform goes beyond chat by handling files, coding, research, connected services, and multi-step tasks autonomously.
Artificial intelligence tools usually depend on large cloud servers, but Perplexity now wants powerful AI agents to run closer to the user. The company has launched Portable Computer, a local AI system built for NVIDIA hardware. The product brings AI models, agent tools, and private workflows onto personal machines instead of sending every task to remote servers.
Portable Computer starts with support for NVIDIA DGX Spark systems and aims to create a new way to use AI. The platform allows users to run AI agents that can handle files, analyze information, write code, and complete tasks with local computer power.
Perplexity Portable Computer works as a complete AI agent platform rather than a simple chatbot. The system combines local models, task planning, file search, tool connections, and secure code execution in one package.
The main idea behind the product focuses on privacy and control. Local processing keeps sensitive files and personal information on the user’s machine. The system can also connect with cloud AI services when a task needs stronger models, but it asks for permission before sending information outside the device.
This approach gives users more control over their data while still allowing access to advanced AI features. Portable Computer creates a middle path between fully local AI and cloud-based assistants.
Perplexity developed Portable Computer with NVIDIA’s local AI hardware ecosystem in mind. The first target platform is NVIDIA DGX Spark, a desktop AI system that uses the GB10 Grace Blackwell chip.
The company also plans support for powerful NVIDIA graphics cards. Systems with GPUs such as the NVIDIA RTX 3090, RTX 4090, and RTX 5090 can support local AI workloads with enough video memory.
The hardware requirements show the current state of local AI. Running advanced models on a personal machine needs strong computing power, especially when users expect fast responses and complex task handling.
Also Read - Perplexity, NVIDIA Bring Local AI Tools for Better Data Privacy
Portable Computer supports local AI models that can work without constant cloud access. The launch includes Qwen 3.8 27B and PPLX 27B, a Perplexity version based on Qwen.
These models allow the system to complete many AI tasks directly on local hardware. Users can analyze documents, search through files, review information, and manage workflows without sending every request to outside servers.
Perplexity also plans to add NVIDIA Nemotron 3.5 Lightning in the future. This expansion could bring more options for users who want different models for different tasks.
The platform focuses on practical tasks rather than only conversations. Portable Computer can connect with services such as Google Drive, Gmail, Slack, and GitHub.
A user can ask the AI agent to review documents, examine code, organize information, or prepare reports. The system can complete several steps on its own instead of waiting for a new instruction after every action.
This agent-based approach separates Portable Computer from traditional AI chat tools. The system aims to act more like a digital assistant that manages tasks instead of only answering questions.
Perplexity does not position Portable Computer as a replacement for cloud AI. Instead, the company combines local processing with optional cloud support.
Simple and private tasks can stay on the device. More difficult requests can move to stronger cloud models after user approval. This design gives users flexibility while keeping control over important information.
The hybrid model also solves a major challenge in AI. Local systems offer privacy and ownership, but cloud services often provide stronger performance. Portable Computer attempts to combine both advantages.
Portable Computer represents a larger shift toward personal AI systems. Companies across the technology industry now explore ways to bring advanced AI tools from data centers to personal devices.
For NVIDIA, the partnership creates another use case for AI-focused hardware. More users running local AI agents could increase demand for powerful desktop systems and high-end graphics cards.
For Perplexity, the product expands its role from an AI search company into a provider of AI tools that can operate across different environments.
Also Read - How to Master Perplexity AI for Research Like a Pro in 2026
Portable Computer shows how AI assistants may change from simple question-answer systems into task-focused digital workers. The success of this approach will depend on hardware costs, model quality, and ease of use.
The launch also highlights an important shift in AI development. More users want powerful tools that respect privacy and offer greater control over personal data. Perplexity’s new platform places local computing at the center of that movement and gives NVIDIA hardware a direct role in the next stage of AI adoption.
1. What is Perplexity Portable Computer?
It is a local AI agent platform designed to run AI models, tools, and workflows on compatible personal NVIDIA hardware.
2. Which NVIDIA hardware does it support?
The initial platform targets NVIDIA DGX Spark, with support intended for systems using GPUs such as the RTX 3090, RTX 4090, and RTX 5090.
3. Can Portable Computer work without cloud AI?
Yes. Supported local models can handle many tasks directly on the user's machine, reducing dependence on remote AI services.
4. What can the AI agents do?
Agents can analyze files, search information, work with code, manage documents, and connect with services such as Google Drive, Gmail, Slack, and GitHub.
5. Why is local AI important?
Local AI can provide greater privacy and control over sensitive information while reducing reliance on cloud infrastructure, although powerful hardware is required for advanced workloads.
