The world of AI is moving faster than ever, shifting from tools that answer your questions to autonomous agents that do the work for you. Perplexity, known for its answer engine, is now taking a major step into this new frontier.
What Are These New 'Work Agents'?
Perplexity has launched
a new feature called 'Teammates,' which are essentially AI work agents designed for the enterprise market. Unlike a typical chatbot session that ends when you close the tab, these Teammates are 'always-on'. This means they are persistent, capable of handling long-running, complex projects with minimal human supervision. You can assign them specific roles, such as 'Marketing Researcher' or 'QA Engineer', and they will work continuously in the background to achieve the goals you set. The system is designed for professional teams in areas like sales, operations, and customer support, putting Perplexity in direct competition with similar workplace offerings from giants like OpenAI and Anthropic.
The Engine Behind the Agents: Perplexity Computer
These Teammates don't operate in a vacuum; they are built on top of the 'Perplexity Computer' system. Think of Perplexity Computer as the central engine or operating system that allows AI to execute multi-step tasks. It's a cloud-based platform that can break down a large objective into smaller sub-tasks, coordinate different specialized AI models, browse the web, access files, and use software tools. The Teammates feature adds a new layer of capability to this system, focusing it on sustained, large-scale work rather than just one-off requests. Each agent runs in its own sandboxed virtual environment, complete with a browser and file system, to perform its work safely and independently.
From 'Answering' to 'Doing'
This launch marks a significant strategic shift for Perplexity. The company initially gained fame as a conversational 'answer engine' that provided clear, sourced summaries to user questions. Now, it's moving from simply providing information to autonomously taking action. Instead of asking a question and getting a response, a user can now describe an end goal—like comparing three vendors or monitoring a project's progress—and the Teammate agent will figure out the steps and carry them out. These agents can even work with each other, manage new agents as workloads increase, and proactively monitor for changes, alerting a human user only when a key decision is needed.
How It Works in Practice
A practical example might involve assigning a Teammate to research a competitor. The agent could continuously scan for news, analyze their product updates, and compile a weekly report. In another case, a sales_focused Teammate could identify customers with declining engagement, investigate their support ticket history, and draft a personalized outreach offer for a human to review and send. Users can track the progress of these agents through dedicated threads in the Perplexity app or receive updates via Slack. This functionality comes at a premium, as it is available exclusively to subscribers of Perplexity's 'Max' plan, which costs $200 per month.
The Future of AI in the Workplace
Perplexity's move is part of a broader industry trend toward 'agentic AI'—systems that act as digital proxies for human workers. The goal is to delegate significant portions of digital labor, freeing up employees to focus on strategy, final approvals, and creative problem solving. However, this evolution is not without challenges. The introduction of persistent, always on agents that access company data raises important questions about data governance, security, and the potential for errors to be 'sticky' if an agent learns the wrong information. As these digital teammates become more common, organizations will need to balance the promise of massive productivity gains with the need for robust oversight and control.
