TL;DR: Perplexity released a fully local version of its agentic computer runtime, running entirely on NVIDIA DGX Spark hardware with no cloud dependency.
Summary: Perplexity announced Portable Computer, a fully local deployment of Perplexity Computer built for NVIDIA DGX Spark. The entire runtime—including the orchestrator LLM, subagent LLM, and agent harness—runs on local hardware, removing cloud dependency. This targets developers and enterprises needing private, on-premises AI agent execution.
Why it matters: AI builders get a reference path for running complex multi-agent orchestration entirely on edge hardware, which is critical for privacy-sensitive and offline use cases. Try prototyping agent workflows on DGX Spark with Portable Computer to evaluate local-only latency and data control.
Source: x_com