Perplexity unveils a Computer feature that splits tasks across local models and cloud-based models, to keep private data on-device and maximize token efficiency
Context & Ripple Effects
Perplexity’s Computer product was introduced as a general-purpose digital worker that can route work across multiple AI models, initially for Max subscribers. It was subsequently extended through a macOS-focused Personal Computer offering and an enterprise version.
This update adds a local-versus-cloud routing layer to that product arc. It matters because the product is moving from model selection alone toward deciding where AI work should execute, with privacy and token use becoming part of the agent’s operating logic.
First-order effects
- Perplexity Computer can keep tasks involving private data on-device while sending other work to cloud-based models, changing how its users’ work is distributed across local and remote compute.
- Perplexity gains a more differentiated control layer for its agent product: routing is now framed around both data handling and token efficiency, not only access to many models.
Second-order effects
- Enterprise buyers evaluating Perplexity’s cloud offering may have a clearer path to use agent workflows for sensitive work, while retaining cloud models for tasks that benefit from them.
- Competing agent and AI-assistant products face pressure to make their own local/cloud boundaries legible, since model quality alone does not address where user data is processed or how usage is optimized.
Third-order effects
- If hybrid routing becomes a standard agent capability, AI products may increasingly compete on orchestration policies—what stays local, what goes to the cloud, and how that choice is exposed to users—rather than on a single underlying model.
- The pattern could shift value toward platforms that combine device-level access with broad cloud-model availability; its durability will depend on whether local models can handle enough useful work without eroding the quality advantage of cloud systems.
The trend: This is one data point in the shift from standalone AI models toward hybrid agents that orchestrate models and compute locations according to privacy, capability, and usage constraints.