The world of AI assistants is moving beyond single questions and simple commands. Perplexity's new 'Teammates' feature introduces a major shift, offering a team of AI agents that can own ongoing responsibilities and manage tasks in parallel.
What Are Perplexity Teammates?
Imagine hiring
a team of digital specialists, assigning them each a core responsibility, and having them work around the clock without needing constant direction. That's the core idea behind Perplexity Teammates. Announced in early November 2026, this feature transforms the AI from a reactive chatbot into a proactive, persistent team of agents. Instead of giving the AI a single prompt like 'summarize this report', you assign a Teammate an ongoing role, such as 'monitor all incoming customer support tickets' or 'track our main competitors' market activity'. These AI agents then own that area of work, building on previous tasks, remembering project goals, and even collaborating with each other. The feature is an extension of Perplexity Computer, the company’s existing platform for handling multi-step tasks, but elevates it to a new level of autonomous project management.
The Power of Parallel Work
The most significant innovation highlighted in the headline is the ability for these AI agents to work simultaneously on different jobs. When a user gives a Teammate a large, complex goal, the AI intelligently breaks it down into smaller, manageable tasks. It then identifies which of these tasks can be performed in parallel and spins up additional agents to handle them, with each agent operating in its own isolated environment. For example, one Teammate could be tasked with finding bugs on a company website, and as it discovers them, it can pass those findings to another Teammate whose job is to draft the necessary code fixes. This multi-agent orchestration means projects don't have to be completed in a slow, linear fashion. Multiple streams of work can progress at once, with the primary Teammate coordinating the effort, checking the results, and presenting a consolidated final report or outcome for human review.
How It Changes the Workflow
This new capability signals a move from active prompting to passive delegation. For businesses and professionals, this means a significant reduction in the time spent on coordination and routine follow-ups. Teammates are designed to be proactive. They can monitor connected workplace tools like Slack, GitHub, and Gmail, and initiate tasks automatically when something changes—like a new bug report being filed or an urgent email arriving. A user could assign a Teammate to a Slack channel, allowing human colleagues to hand off work or ask for status updates directly. For instance, an operations-focused Teammate could be told that a certain vendor needs two weeks' notice and will automatically factor that lead time into all future project schedules it manages. This creates a system where the AI handles the repetitive, process-driven work, notifying its human counterpart only when key decisions or approvals are needed.
The Bigger Picture and Current Limitations
Perplexity's Teammates feature is part of a broader industry trend toward creating persistent, autonomous AI agents that function more like digital employees than simple tools. The goal is to offload significant chunks of cognitive labour, from market research and data analysis to managing complex operational workflows. However, this advanced capability is currently positioned as a premium offering. It is available as a research preview exclusively for Perplexity Max subscribers, a plan which comes at a significant cost. While the company is offering a promotional period with free usage, the long term credit consumption of having multiple agents working 24/7 is a practical consideration for potential users. The initial rollout is for the web, with mobile access planned for the future. As with any powerful new technology, there are also questions of trust and oversight, though Perplexity emphasizes that users can always review progress, redirect agents, and take over tasks when needed.
