Perplexity Wants AI to Be Your Colleague: Handle Tasks in Parallel and Remember All Your Previous Projects

As reported by The Verge, Perplexity has released Teammates, an AI agent built specifically for work scenarios. Its positioning is very straightforward: it is fighting to capture the AI agent entry point in daily workflows against OpenAI and Anthropic.

Teammates are resident persistent agents that can take on multiple distinct roles in a single workspace. They will bring all your existing projects, skills and context saved in Perplexity tools over to new tasks, which reduces the redundant cost of repeatedly restating background context for users.

The product is integrated with Perplexity Computer, the feature that deploys agents on your device to complete assigned tasks. According to Perplexity, Teammates adds collaborative capabilities on top of Perplexity Computer, as well as the ability to carry out sustained, large-scale work. You can assign different work to each Teammate and have them run tasks at the same time. The Verge compares this product form to OpenAI's Dots and Anthropic's Claude Duet.

The agents described here can be understood as AIs that can take on tasks, call integrated tools, and push work forward continuously. Different from one-off Q&A interactions, their tasks can span multiple steps, cross different projects, and even require multiple roles to run in parallel. Perplexity started as an AI search provider, and this new push to move its products into workflows centers its competitive edge on memory retention, project context awareness, and parallel task division.

From available public information, users assign roles and tasks to Teammates in the workspace, and the agents will access projects, skills and saved memory from your existing Perplexity tools. The original article does not elaborate on specific access entrances, permission boundaries and billing methods.

The product direction is very clear: AI assistants are shifting from simply answering questions to taking full ownership of tasks. Perplexity uses multiple persistent resident agents to cover continuous ongoing work. Whether this direction can be successfully validated depends on whether it can reliably handle real-world project context and support stable collaboration.

Source: The Verge, Perplexity is taking on OpenAI and Anthropic with work-focused AI agents