Perplexity launches persistent AI agents for Max subscribers

From queries to coworkers

Perplexity has officially launched Teammates, a new suite of personal AI agents designed to handle ongoing work. Unlike standard chatbots that require new instructions for every interaction, these agents are built to take over responsibilities based on high-level goals. They function as persistent coworkers that stay active across multiple tasks.

The company differentiates these from its previous 'Computer' tool, which operated in temporary sandbox sessions that vanished after each job. Teammates are designed to stick around, maintaining a continuous presence in a user's workflow. This shift aims to move AI from a simple search tool to a functional member of a digital team.

CEO Aravind Srinivas describes the new feature as coworkers that can complete tasks from start to finish. By retaining context, the agents reduce the need for users to constantly re-explain their requirements. This persistence is the core differentiator for the new offering.

How the agents learn and act

Teammates gain their capabilities by accessing a user's specific data, including Projects, Memory, Skills, and various file connectors. They also draw on information from past sessions to understand the nuances of a user's specific work style. This allows them to handle a wide variety of professional use cases across different departments.

The technology is capable of performing complex operations like analyzing data, monitoring security, and running quality assurance loops. For example, in a marketing context, a Teammate can identify a successful ad campaign, prepare a larger version of it, and send a contract extension to the legal team. It can also manage A/B tests and triage technical bugs independently.

A key feature of these agents is their ability to learn from human feedback and mistakes. If a brand director tells a Teammate to perform a specific action every time a certain metric is met, the agent remembers that instruction for the future. This creates a loop where the AI becomes more aligned with the user's preferences over time.

Integration and accessibility

Users can currently access Teammates on the web through the Perplexity platform. The company plans to expand this functionality to native mobile and desktop applications in the near future. This expansion will allow for more seamless integration into the daily habits of power users.

The service is not available to the general public for free. To use Teammates, users must upgrade to the Max tier, which carries a significant price tag. As digitaltrends.com reported, the cost for this subscription is $200 per month.

The platform also offers specific ways to interact with these agents in common work environments. For instance, users can add a Teammate to a Slack channel where it can watch the conversation and act when specific programmed changes occur. Users maintain oversight by being able to review the agent's work, take over its desktop, or redirect its actions at any time.

The competitive landscape

Perplexity is entering a crowded market of AI agents that aim to automate professional workflows. Several major tech companies are already developing similar technologies to compete for the same space. These include OpenAI's workspace agents and Microsoft's Copilot Cowork, though those currently target organizations rather than individuals.

Other notable competitors include Grok Bot and Anthropic's Claude Cowork. While many of these tools are emerging simultaneously, Perplexity is positioning its Teammates as a persistent solution for individual subscribers. The race is moving toward creating agents that can operate autonomously within specific environments.

The focus for these companies is increasingly on 'end-to-end' task completion. Instead of just generating text or images, the goal is for the AI to navigate software, manage communications, and execute multi-step projects. Perplexity's move suggests that the next phase of AI utility is defined by persistence and autonomy.

What to expect next

The rollout of Teammates represents a significant step toward the automation of routine business operations. As the agents become more integrated into mobile and desktop environments, they may become more central to how individuals manage their daily to-do lists. The ability for an AI to 'learn' from a single instruction and apply it forever is the primary goal here.

For users, the high cost of the Max tier means these tools are currently positioned for high-value professional work. However, as the technology matures, the barrier to entry for these types of persistent agents may decrease. For now, the focus remains on high-level goals like marketing, sales, and operations.

The success of Teammates will likely depend on how well the agents can handle complex, multi-step instructions without human intervention. As users provide more feedback and the agents refine their skills, the distinction between a tool and a coworker may continue to blur. Watch for how these agents handle increasingly complex data connectors and real-time security monitoring.