minssam.
Published on

Perplexity Teammates: AI Agents That Keep Working While You Sleep

AI agents have officially been given a new job title: teammate.

On November 5, 2026, Perplexity announced a new feature called Teammates. The name is deliberate β€” not a tool, not an assistant, but a colleague. And the design backs up the branding.

Perplexity Teammates keep working after you close your laptop. They collaborate with other AI agents, and they can message real human coworkers on Slack. This is one of the most direct implementations of the "AI as coworker" concept on the market today.

What Makes This Different from AI Assistants

Until now, AI assistants worked on a request-response loop. You type, the AI replies, you type again, the AI replies. When the conversation ends, the AI stops.

Teammates change this model. You give them a high-level goal, and they break it into smaller tasks, route each to the most suitable AI model, collect results, and move to the next step β€” all without waiting for another prompt.

This runs on Perplexity Computer, a system that automatically selects from a pool of 19 AI models based on the task at hand. Writing tasks get a different model than coding tasks or research tasks. Each unit of work gets the right tool.

Memory and Permissions

Teammates inherit the user''s existing Perplexity environment. Projects, personal Skills, memory, and tool permissions carry over automatically. There is no need to re-introduce yourself or re-explain your context from scratch.

The Slack integration stands out. When a Teammate reaches a decision point that requires human judgment, it sends a message to a Slack channel. When a human team member responds, the Teammate picks up and continues. The result is a workflow where AI and humans collaborate inside Slack as equals β€” each handling the parts they''re best suited for.

What Roles Can a Teammate Fill?

Perplexity names marketing, sales, customer support, and operations as target use cases. More concretely, it lists roles like Engineer, QA Engineer, and Legal Research Assistant as examples you can configure.

Translated into an edtech context, here are some practical possibilities:

  • Content Curator Agent: Collects daily education news, ranks items by relevance, and drafts a newsletter ready for review each morning.
  • Learning Analytics Agent: Monitors student activity data periodically and sends a Slack alert to the responsible teacher when early dropout signals appear.
  • Sales Support Agent: When a potential client submits an inquiry, it researches the organization and delivers a tailored proposal draft to the account manager.

Multiple Teammates can run simultaneously, each with its own role, only surfacing to humans when a decision is required.

Pricing and Access: The $200 Reality

Teammates are currently available as a research preview for individual Max plan subscribers. Max costs 200permonth,or200 per month, or 2,000 per year. At launch, Perplexity is offering free Teammates usage included with Max as a promotional offer β€” but a separate pricing tier for Teammates seems likely long-term.

Access is currently limited to individual subscribers on web and iOS. Team and enterprise plans are in development.

The price is real. But the ROI calculation shifts if Teammates genuinely replace hours of repetitive work per week. Perplexity''s pitch to enterprise buyers is essentially: one Teammate costs less per month than one hour of a full-time employee''s time per day.

What This Moment Tells Us About AI Work

The arrival of Teammates is not just a product launch. It is a signal that AI is shifting from "tool you use" to "colleague you direct."

The implication for skills: the most valuable human capability in an agentic workflow is not the ability to do the task yourself, but the ability to define the goal clearly, evaluate the agent''s output, and decide when to hand control back to a human. That metacognitive layer β€” goal-setting, review, judgment β€” is where human contribution concentrates.

For students entering the workforce in the next decade, fluency with AI agents will be as expected as email fluency is today.

Tips for Getting Started

  • Assign simple, repetitive tasks first: Anything with a clear pattern and a known output format is an ideal first Teammate assignment.
  • Connect Slack before anything else: The Slack integration is where Teammates show their real value. Without it, you lose half the experience.
  • Give high-level goals, not step-by-step instructions: "Analyze competitor marketing for the Q4 report" outperforms "summarize these three files."
  • Build a daily log review routine: Even with autonomous agents, checking the work log once a day is non-negotiable.
  • Treat this as a research preview: Impressive for early access β€” but keep critical decisions in human hands for now.

Sources:

Perplexity Teammates: AI Agents That Keep Working While You Sleep | MINSSAM.COM