Perplexity Teammates Launches Always-On AI Agents for Enterprise Work

Perplexity has introduced Teammates, a new class of always-on AI agents designed to operate continuously within enterprise workspaces. Unlike traditional chat assistants, these agents can execute multi-step tasks, retain context across sessions, and collaborate on long-running projects.

The system builds on Perplexity Computer, which deploys agents locally on user devices. Teammates extend this capability by enabling persistent, large-scale workflows that run 24/7. Each agent receives a dedicated thread for monitoring and can be accessed via Slack integration.

Access requires Perplexity's $200-per-month Max plan, positioning the feature at the premium end of the market. CEO Aravind Srinivas has stated a vision where every organization deploys hundreds of such agents per employee.

Why it matters for GPU infrastructure

Continuous agent operation translates to sustained GPU demand. Unlike bursty inference workloads, always-on agents require persistent compute allocation, high memory capacity, and low-latency networking. This trend accelerates the need for scalable GPU cloud platforms that can provision dedicated instances on demand.

For AI infrastructure providers, the rise of agentic workflows signals a shift from intermittent inference to continuous, multi-agent orchestration. Capacity planning must account for concurrent agent threads, each consuming GPU resources for reasoning, tool use, and memory retrieval.