Perplexity Launches 'Personal Computer' to Orchestrate AI Tasks

AI search company Perplexity has launched a new orchestration software, dubbed 'Personal Computer,' designed to manage desktop files and applications. The system intelligently distributes AI workloads between a user's local PC and cloud servers to optimize for cost and efficiency, routing simpler tasks locally and more complex ones to the cloud.
Perplexity Launches 'Personal Computer' to Orchestrate AI Tasks

Perplexity Launches ‘Personal Computer’ to Orchestrate AI Tasks Perplexity is betting that the future of AI won’t live solely in distant data centers, but will be choreographed across the laptop on your desk and servers in the cloud.

Laying the groundwork: From browser to “Computer”

Perplexity first pushed beyond search with Comet, an AI-powered browser launched in July 2025 to automate background research and task execution. By October 2025, Comet became freely available worldwide on desktop, attracting enterprise users through deals like a Coinbase partnership that piped real‑time crypto data directly into the browser.

In May 2026, Perplexity embedded its new orchestration layer, called Computer, into Microsoft 365 in two phases: first inside Microsoft Teams on 4 May, then as native side panels in Word, Excel, PowerPoint and Outlook on 28 May. The system can “actively pilot desktop files and local apps,” coordinating multi‑step projects across corporate documents, spreadsheets and email.

Hybrid inference: Splitting work between PC and cloud

At Computex in early June, CEO Aravind Srinivas unveiled a platform that “dynamically splits AI workloads between personal computers and cloud servers,” acting as an “air-traffic controller for AI tasks” to cut inference costs. Simple jobs like summarisation run on the user’s PC, while complex reasoning is routed to cloud GPUs, with decisions made in real time.

Intel highlighted the demo as “keeping sensitive data on device while cloud AI adds scale and context,” underscoring the appeal of hybrid orchestration for privacy‑conscious enterprises. Perplexity echoed this, noting that Computer can “split tasks between a local model running on your machine and frontier models in the cloud,” keeping private data local while maximising “token efficiency.”

The “value per watt” thesis and a new “Personal Computer”

Srinivas frames the strategy around “value per watt per user,” arguing that winners will balance “accuracy, latency, cost, privacy and intelligence” rather than just chase ever‑larger models. To operationalise that, Perplexity is rolling out Personal Computer, an orchestrator that unifies models, apps and hardware across Windows and Mac by routing tasks through “the most efficient path” among local files, desktop applications and cloud services.

Industry voices see this hybrid, multi‑model routing as part of a broader shift: using open and proprietary models together to optimise cost and quality, and moving more inference to the edge while cloud systems add scale and context.

Continue reading https://foxvector.com/stories/019e93a0-4d5a-2bb1-711e-1133a1c5afe4

Write a comment
No comments yet.