The biggest headache with AI assistants has been their short memory. You ask, they answer, and the next chat starts from zero. Perplexity is aiming to fix that with a feature that transforms its AI from a query machine into a persistent employee.
What Are AI Teammates?
Perplexity
has launched a new feature called 'Teammates', a system of AI agents designed to take ownership of ongoing work. Instead of just responding to one-off questions, users can assign a 'Teammate' a long-term responsibility, much like delegating a project to a human colleague. Announced on November 4, 2026, this capability allows the AI to manage complex projects, monitor developments, and follow through on tasks without needing to be prompted again and again. Think of it as the difference between asking a stranger for directions and hiring a personal assistant who learns your goals, remembers your projects, and actively works on them in the background.
From One-Off Prompts to Ongoing Tasks
The core innovation here is the shift from a reactive to a proactive AI. Traditionally, interacting with an AI has been a series of isolated conversations. You might ask it to summarize an article, write an email, or analyze data, but each request is a self-contained task. Teammates are built for persistence. You can assign one a role, such as 'market researcher' or 'social media monitor'. That Teammate will then continuously work on that assignment, retaining the context, files, and instructions from previous interactions. For example, a Teammate could be tasked with tracking competitor pricing and delivering a weekly summary, a job it performs autonomously until you change its role. This is a significant step toward AI agents that function less like a search engine and more like a digital employee.
How It Actually Works
Teammates build on Perplexity's existing 'Computer' platform, which already allows AI to use tools and code. What's new is the 'multi-agent orchestration'. When you give a Teammate a large goal, it can break the work into smaller sub-tasks and assign those to other, newly created agents that run in parallel. Each agent operates in its own sandboxed environment with its own desktop and browser. Teammates can connect to workplace tools like Slack, GitHub, and Gmail, allowing them to monitor channels, review code, or triage support tickets. For instance, a support agent could identify recurring bug reports, while another agent takes those findings to draft potential fixes. The system is designed to notify the user when a decision is needed, but otherwise handles the coordination itself.
The Competitive Landscape and Cost
Perplexity is not alone in the race to build more capable AI agents. Companies like OpenAI and Anthropic are developing similar work_focused assistants. However, Perplexity's approach is distinctly centered on assigning persistent roles to agents that can collaborate with each other and even with human colleagues via integrations like Slack. This functionality comes at a premium. The Teammates feature is initially available as a research preview exclusively for Perplexity Max subscribers, a plan that costs around $200 per month. While Perplexity is offering a promotional period with free usage, it will eventually draw from a user's monthly credit allotment, making it a tool geared toward professionals and businesses that can justify the cost for automating complex workflows.
What This Means for the Future of Work
The introduction of AI Teammates signals a broader trend in the AI industry: the move from simple chatbots to autonomous agents. The goal is to create systems that can reliably manage entire workstreams, freeing up human workers to focus on strategy and outcomes rather than the minutiae of task coordination. Perplexity's CEO has described a future where a company might have hundreds of AI teammates working for each employee. While the technology is still in its early stages and requires human supervision, it paints a clear picture of where AI is headed. We are moving from tools that answer our questions to partners that manage our projects.
