Perplexity Launches Persistent Teammate Agents as AI Workflows Move Beyond Simple Copilots

Perplexity has launched Teammates, persistent AI agents designed to handle complex, ongoing workplace assignments with limited supervision. The agents live inside Perplexity’s product environment and can draw on projects, skills, memory, and work history rather than starting each interaction from scratch.

The launch places Perplexity alongside OpenAI, Anthropic, and Microsoft in a rapidly expanding race to build AI systems that execute work over time. Teammates are available to Perplexity Max subscribers, whose plan costs $200 per month, according to Engadget. The company’s own product pages describe them as agents that “work for you,” while its launch material positions them as purpose-built extensions of Perplexity Computer.

Teammates remain available in dedicated threads after users create them, allowing people to return to an agent rather than recreate instructions for every assignment. Each agent can be configured around a role or recurring responsibility, with access to the context accumulated through Perplexity’s projects and related tools.

The design separates Teammates from one-off prompts and conventional chat sessions. Perplexity Computer already handles agentic workflows such as research and task execution, while Teammates are intended to maintain continuity across a longer-running stream of work. The company’s interface includes a dedicated Teammates area and a “Create a teammate” entry point, according to Perplexity’s product interface and help materials.

Persistent context is the central value proposition. An agent that remembers a project’s goals, prior decisions, preferred tools, and accumulated work can spend less time rebuilding background and more time completing the next step.

That model targets work that resists simple automation: coordinating operations, producing marketing deliverables, monitoring projects, and following a process across multiple applications. Perplexity’s launch framing connects Teammates to the company’s earlier work with Comet and Computer, while Asana’s separate AI Teammates announcement reflects the same shift toward agents that collaborate with humans on nuanced workflows rather than merely completing isolated commands. Asana described its agents as collaborators intended to improve work velocity and quality.

Perplexity’s timing reflects a broader change in how major AI companies position agents. OpenAI and Anthropic have introduced systems aimed at completing multi-step work, and Microsoft has embedded agent capabilities into workplace software. Perplexity is betting that its advantage comes from combining web research, Computer’s execution features, and long-term project context in one environment.

The competitive question is whether users will pay for a persistent agent because it saves meaningful time, rather than because it offers another chat interface. The $200 monthly Max price sets a high bar, especially when companies can also evaluate agents built into productivity suites. A separate startup called Teammates described agents that operate through Slack, Teams, email, and thousands of applications, showing how quickly the “AI coworker” concept is spreading beyond search and browser tools.

Persistent agents also create operational responsibilities. The longer an agent retains context and acts across tools, the more important permissions, review points, data boundaries, and error recovery become. A mistaken one-off answer is easier to catch than an incorrect action repeated across an ongoing project.

Perplexity’s available descriptions emphasize autonomy and continuity, but the launch materials provide limited detail about governance controls, audit trails, or how users approve sensitive actions. Those details will shape adoption in businesses where agents touch customer data, internal documents, financial systems, or external communications. The broader market’s promise of agents behaving like coworkers, including products that receive company email addresses and work inside collaboration platforms, raises the same questions about identity and accountability.

Perplexity’s next test is sustained performance across real projects. Teammates need to preserve useful context, select the right tools, recover from failures, and produce work that remains accurate after several handoffs and revisions.

Success would give Perplexity a stronger role in workplace software and turn Computer from an on-demand tool into a home for recurring digital labor. Failure would leave users with expensive agents that require the same close supervision as ordinary chatbots. Asana’s collaboration-focused announcement and the separate Teammates startup’s cross-application approach show that the market is converging on a common goal: AI systems that participate in work continuously, while companies compete over trust, integrations, and measurable results.