Perplexity announced the launch of Teammates, a system of artificial intelligence (AI) agents designed to take on ongoing tasks in work environments. The feature allows different responsibilities to be assigned to assistants that operate simultaneously, track projects, and can act when they identify changes — without needing to receive new instructions at every step.
The feature was introduced this Thursday (5) and works integrated with Perplexity Computer, the company's platform that uses agents to perform tasks. The difference is that Teammates was developed to keep activities underway, organize larger-scale work, and collaborate with other agents and employees.
Each agent can take advantage of projects, skills, memories, files, and tools already configured by the user in Perplexity. The idea is to avoid having to explain the context of an activity again every time a new task begins.
The feature is initially available on the web for subscribers to the Max plan, which costs US$ 200 per month (around R$ 1,092). Native apps for mobile devices and computers are part of the company's plans, but have not yet been released.
How do Teammates agents work?
- Users can create different agents and assign a specific role to each one. They can work in parallel, track activities, and share tasks with one another.They can also interact with employees through Slack, a corporate communication platform;
- In practice, one agent may be responsible for tracking metrics from a marketing campaign, while another monitors sales activities or organizes customer service requests;
- Perplexity also cites applications in operations and software development. Possibilities include tracking deadlines, analyzing data, identifying problems, conducting quality tests, and monitoring security issues;
- Each Teammate operates in its own workspace, with separate computer and browser sessions. The user can monitor the progress of activities, review what has been produced, take control of the environment, or redirect the agent when necessary;
- The Slack integration allows colleagues to assign tasks to agents as well. This way, they can monitor work channels and act as conversations and connected projects change.
Now, multiple AI agents can work for companies as coworkers – Image: Disclosure/Perplexity
Read more:
- ‘Hey Plex’: Samsung puts Perplexity into Galaxy AI
- Personal Computer: Perplexity’s AI that works on its own on your PC day and night
- Folha de S.Paulo files lawsuit against Perplexity
Feature costs R$ 1,092 per month
Teammates requires a Perplexity Max subscription, the company’s most advanced plan, which costs US$ 200 per month (approximately R$ 1,092).The option can also be contracted for US$2,000 per year (around R$10,924).
The new feature expands the capabilities of Perplexity Computer, which already allows the use of different AI models and tools to perform more complex tasks. With Teammates, the company seeks to turn these features into assistants that take on ongoing responsibilities rather than merely responding to one-off requests.
Perplexity CEO Aravind Srinivas said the company’s vision is for organizations to have hundreds of agents working for each employee. “Over time, we imagine that every organization will have hundreds of teammates working for each person,” the executive wrote in a post on X.
Perplexity enters the race for AI agents
The launch places Perplexity in an increasingly competitive race to develop AI agents focused on work.
OpenAI introduced Dots in September, featuring assistants that can operate continuously, learn from users and perform tasks through their own computer in the cloud. Anthropic, in turn, opened a research preview of Claude Duet in October, a tool that also uses proactive agents for professional activities.
These initiatives represent a change in how AI is used. Instead of relying solely on questions and answers, companies are developing systems capable of receiving objectives, planning steps, using tools and monitoring tasks over extended periods.
Perplexity’s proposal is to bring this model to different roles within an organization, with agents that share information and collaborate with one another.
Rodrigo Mozelli is a journalist graduated from São Paulo Methodist University (UMESP) and currently works as a writer for Olhar Digital.
