Realtime AI News
Perplexity's Portable Computer Lands on Windows, Powered by NVIDIA RTX
On September 14 NVIDIA said Perplexity is bringing Portable Computer, the local version of its Perplexity Computer agent, to the Perplexity app for Windows on compatible GeForce RTX PCs and RTX PRO workstations. It runs multistep tasks on local models, keeps sensitive files on the device, and does not spend Perplexity Computer credits for work completed locally.
On September 14, NVIDIA's blog announced that Perplexity's Portable Computer is now available in the Perplexity app for Windows, on compatible NVIDIA GeForce RTX PCs and RTX PRO workstations. The release builds on earlier support for NVIDIA DGX Spark systems and RTX PCs running Linux.
Portable Computer is the local version of Perplexity Computer, the company's cloud agent. It plans and carries out multistep tasks, using local models to analyze data, bring together information across files, and handle recurring work.
The pitch is that sensitive information stays on device, and work completed locally does not consume Perplexity Computer credits. When a task calls for more advanced research or reasoning, Portable Computer can identify that and ask the user for permission before sending information off-device.
Setup is deliberately simplified. The app ships with a local model, such as Qwen 3.8 27B, that is post-trained to work with Perplexity Computer and optimized for NVIDIA RTX GPUs, so users do not have to research models or configure a complex local software stack.
Local and cloud AI come together in one app: Computer's built-in browser and proprietary SPACE sandbox are available, and connectors for Microsoft Outlook, OneDrive, Word, Google Drive, Gmail, Slack and GitHub pull the files and apps already in users' daily workflows into the agent.
Portable Computer requires NVIDIA GeForce RTX or RTX PRO GPUs with 24GB or more of VRAM, and NVIDIA says DGX Station support is expected to come soon. The practical effect is that 27B-class models can now run on a wide range of consumer and professional cards instead of only in the cloud.
What matters here is the cost structure of a local agent: inference happens on the machine, cloud credits are spent only where they are needed, and a company's most sensitive files never leave the endpoint. For compliance-bound teams, a default-local and cloud-when-needed agent is an easier sell than a purely cloud-based one.
Two things to watch next: how well the local models complete real tasks and where their limits sit, meaning which jobs still require the cloud; and how quickly support reaches DGX Station and other platforms, along with whether the connector list keeps expanding into more enterprise systems.
Why it matters
Perplexity and NVIDIA are extending the cloud agent business model onto the PC itself, keeping sensitive data and part of the inference bill on the device. For developers and compliance-sensitive enterprises, a local Windows agent now becomes a default option worth evaluating rather than a niche experiment.
Nearby Updates
All09/14, 22:54
Oracle Health pitches a clinical AI agent aimed at nurses' documentation burden
A September 14 report describes Oracle Health's Clinical AI Agent as a tool that helps nurses alleviate documentation burden and streamline care. It extends clinical AI from physician note-taking into nursing workflows, where the recording load is heavier and more fragmented.
09/14, 22:45
Superhuman acquires YC-backed notetaker Fathom as productivity platforms chase agentic work
Superhuman said on September 14 that it is acquiring Y Combinator-backed meeting notetaker Fathom, giving its productivity suite a direct feed of meeting topics and action items. Fathom reports more than 400,000 monthly active users and says over 1 million people have recorded meetings on the platform.
09/14, 22:05
NVIDIA's Groq 3 LPX enters full production, promising the world's fastest inference
The Futurum Group reports that NVIDIA's Groq 3 LPX has entered full production, with the claim of delivering the world's fastest inference as its core pitch. So far the report offers only the production milestone and the performance claim, with deployment customers, availability regions, pricing and benchmark conditions still undisclosed.
09/15, 00:00
Anthropic CEO Dario Amodei Calls for an AI Slowdown
The New York Times reports that Anthropic CEO Dario Amodei has publicly called for slowing the pace of AI development. Coming from the head of one of the leading frontier labs, the statement sharpens the tension between safety messaging and racing competition.