Perplexity announced on Thursday (November 5, 2026) Perplexity Teammates, AI agents designed to act as persistent coworkers. Unlike chat sessions that require constant prompting, Perplexity Teammates receive high-level goals and manage tasks end-to-end on an ongoing basis.
What are Perplexity Teammates?
According to CEO Aravind Srinivas, Perplexity Teammates shift the paradigm from “writing prompts in a session” to “giving high-level goals.” They build on Perplexity Computer (the existing multi-model orchestrator), inheriting the user’s projects, memory, skills, and tools. Each Perplexity Teammate has its own desktop, browser, and isolated environment, and can collaborate with other Teammates or human colleagues via Slack.
Use cases cited by the company include bug triaging, QA loops, security monitoring, data analysis, A/B test management, and growth campaigns. Internally, Perplexity says the agents have already saved hundreds of hours of manual work.
Availability and pricing of Perplexity Teammates
Perplexity Teammates are available now to Max plan subscribers ($200/month) on the web. Native mobile and desktop apps are coming soon. For a promotional period, usage is included in the credits already present in the Max plan; afterward it consumes the regular credit allotment.
The approach is more expensive than competitors such as OpenAI Dots or Claude Duet, but Perplexity emphasizes multi-model orchestration (Claude, Gemini, Grok, ChatGPT and others) and the ability to handle continuous large-scale work that Perplexity Teammates offer.
Why Perplexity Teammates matter
The launch of Perplexity Teammates reinforces the trend of AI agents moving from one-off tools to persistent “coworker” roles. With native Slack integration and proactive monitoring, Perplexity Teammates point toward a future in which organizations can have dozens or hundreds of specialized agents per person — something Srinivas has already publicly envisioned.
It is still a research preview. Users can review the work, take over the desktop, or redirect the agent at any time. Real-world effectiveness in diverse corporate environments will be tested in the coming months.
By Grok
