Perplexity launches Teammates for ongoing agent work

Max subscribers can assign persistent agents recurring jobs, including competitor monitoring. Small teams still need to test the review burden.

Perplexity launched Teammates on November 5, introducing persistent AI agents for ongoing work rather than isolated assignments, according to The Verge’s launch coverage.

Perplexity’s changelog lists web access for Max subscribers and describes agents that plan work, run tasks in parallel and prepare results for review. The company says each teammate draws on existing Computer tasks, skills, memory and tools, and can collaborate with other teammates or message coworkers in Slack.

The Decoder reports that each teammate runs on its own virtual machine with a desktop and browser. According to Engadget, the agents run 24/7, can recruit and manage additional teammates, and have individual sidebar threads where users can check status and give new instructions.

For small-business owners evaluating marketing and sales workflows, the company’s examples offer concrete assignments to examine. One asks an agent to monitor competitor price changes, launches and new hires, flag important developments immediately and deliver a weekly summary. These are suggested uses, not evidence of customer results.

A separate Perplexity example covers drafting customer-support replies using policies and order history, with approval required before refunds. PCMag reports advertised roles including Ads Manager, Engineer, QA Engineer and Legal Research Assistant. Those role names describe the offering, not demonstrated performance.

Access and continuing costs also matter for a small team. PCMag lists Max at a $200 monthly subscription or a $2,000 annual subscription. Its launch report says Teammates initially carries no additional charge during a promotional period; subsequent use will require an undisclosed number of credits. That leaves the eventual cost of a recurring assignment uncertain.

A hypothetical trial could stay narrow: monitor a chosen set of competitor changes and prepare a follow-up for human review. An owner could then assess whether the output is useful, how much correction it needs and whether supervision saves time. The same evaluation questions provide a frame for comparison with Anthropic’s offering, discussed in coverage of Claude Duet for small-business workflows.

The launch establishes a proposed workflow and an access tier. It doesn’t establish that persistent agents improve customer acquisition. For businesses deploying agents to find or serve customers, the practical test is whether recurring work produces reviewable output without an excessive burden of checking and correction.

Perplexity launches Teammates for ongoing agent work

Sources