How to Use Perplexity Teammates: Always-On AI Agents for Ongoing Work

For years, using an AI tool has meant the same thing: you ask a question, it answers, and the conversation ends. Perplexity changed that pattern on November 4, 2026, with Teammates — always-on AI agents that take ownership of ongoing work instead of waiting for your next prompt. You give a teammate a job, and it keeps working on it day after day.

This guide explains how to use Perplexity Teammates: what they do, how to set one up, what the Max plan costs, and how to build a similar workflow for less.

What Are Perplexity Teammates?

A Perplexity Teammate is a persistent AI agent assigned to a responsibility. Instead of a chat that forgets everything when you close the tab, a teammate carries its goal, work history, and shared files from one run to the next. It plans tasks, runs them in parallel, monitors connected tools, and acts when something changes — or asks you when it needs input.

A chatbot is a stranger you brief from scratch every time; a teammate is a colleague who already knows the project.

Always-on agents, not one-off chats

The key idea is persistence. Give a teammate a responsibility such as "track our competitors" and it keeps working on it over time, checking connected tools on its own schedule and acting without being prompted. Each teammate runs on its own isolated computer, with its own desktop and browser. You can review what it did, take over its desktop, or redirect it mid-task.

How Teammates differ from Perplexity Computer

Perplexity Computer handles one-off, multi-step tasks inside temporary sandbox sessions: you ask, it does it, and the session ends. Teammates are built for work that never ends.

• Perplexity Computer: one-off multi-step tasks, temporary sessions, no long-term memory.
• Perplexity Teammates: ongoing responsibilities, persistent memory of goals and history, scheduled and event-driven work.
• A regular Perplexity chat: question-and-answer only, no persistence, no independent action.

If your task has a finish line, Computer is the right tool. If it repeats every day or week, that is what Teammates are for.

What Can a Teammate Actually Do?

The launch examples focus on recurring work that quietly eats a freelancer's day. A teammate can triage your inbox and calendar each morning, flagging what needs you and drafting replies for review. It can track competitors — watching pricing pages, launches, and announcements, then summarising what changed — draft support replies, investigate a deal that has gone quiet, and run a research brief on a fixed schedule.

Teammates can also be added to Slack channels, so they can pick up tasks mentioned there and report back where your team already works. They monitor connected tools such as Slack, GitHub, and Linear, and ask you when a decision needs a human — for example, before replying to an important client.

A teammate is best at recurring, well-defined work. The clearer the responsibility, the more useful it becomes — vague instructions produce vague results, just as they would with a new human hire.

How to Use Perplexity Teammates (Step-by-Step)

Teammates are currently in research preview for individual Max subscribers on the web. If you meet those requirements, here is the setup flow.

Step 1 — Check your plan

Confirm you are on the Max plan, which costs $200 per month (or $2,000 per year). On a free or Pro plan, the Teammates option will not appear. Access is web-only at launch, and Enterprise plans are not included by default.

Step 2 — Create your first teammate

In the sidebar, open Teammates and choose Create teammate. Give it a name and a role, then describe the job in plain language. Be specific: instead of "help with email," try "each weekday at 8 a.m., review my inbox, flag client messages, and draft replies for the rest." Include the schedule if the job repeats. The default model is Claude Opus 5.5, changeable per teammate or task. Start with one small, low-risk job, such as a daily competitor summary.

Step 3 — Connect your tools

A teammate is only as useful as the tools it can see. Connect services by asking the teammate directly — for example, "Connect Slack" — and reviewing the permissions it requests. Typical connections include Slack, GitHub, Linear, and Gmail. Saved credentials are encrypted and teammates cannot view them, but still connect only what each teammate genuinely needs, with the narrowest permissions you can.

Step 4 — Set a schedule and review loop

Decide when the teammate works and when you check its output. Build a short daily habit: scan what it did, correct anything wrong, and approve anything it drafted before it goes out. Nothing drafted for a client should go out without a human look.

Step 5 — Give feedback so it learns

Teammates remember goals, history, and shared files across runs, and your corrections are part of that memory. When it drafts a reply in the wrong tone, say so and show the better version. This feedback loop turns a generic agent into one tuned to your business — and it is the step most people skip.

How Much Do Perplexity Teammates Cost?

There is no separate price for Teammates. They are included in the Max plan at $200 per month (or $2,000 per year), and teammates draw on Max's monthly credit allocation. Perplexity has not published a per-task cost, so the real cost of heavy use is not yet public.

Who is it worth it for? If you already pay for Max, Teammates are a genuine addition. If a few hours of triage and drafting each week would free you for billable work, the maths may work. If you only need an agent occasionally, $200 a month is a lot for a feature you will barely use — and the alternatives below will serve you better.

Teammates vs Other AI Agents

Always-on agents are not a Perplexity-only idea. Here is how Teammates compare with the other ways to get agent-style work done.

Option	Persistence	Setup effort	Cost	Best for
Perplexity Teammates	Always-on, remembers goals	Low	Max plan, $200/month	Ongoing research, triage, tracking
Perplexity Computer	One-off sessions	Low	Varies by plan	Single complex tasks
Manus AI	Multi-step task runs	Low	Available on ProAccess.me	General agent work
n8n-style automation	Always-on workflows	High — you build it	Varies by host and plan	Precise, repeatable automations
ChatGPT agents	Session-based with memory	Low to medium	Depends on plan	Chat-first agent tasks

The trade-off is between convenience and control. Teammates are the most convenient persistent agent: describe a job and it starts working. Workflow tools like n8n give you more control, but you build and maintain every step. General agents such as Manus sit in between.

Build Your Own Always-On Agent Workflow for Less

If $200 a month is more than you want to spend, you can assemble much of the same capability from affordable tools. Start with the agent: Manus AI is a general-purpose AI agent that takes on multi-step, ongoing work — the closest affordable analogue to the "give it a job, not a prompt" model behind Teammates. Brief it in plain language and it plans the work and delivers results you can review: weekly research briefs, competitor summaries, and support-reply drafts.

Next, feed the agent fresh data. That is what Apify does: its web scraping and automation actors monitor websites, track competitor pages and prices, and deliver structured data you can hand to an agent. Pair an Apify monitor with a Manus brief and you have a competitor-tracking workflow that runs on a schedule, without a Max subscription.

Finally, keep the running cost low. MUSE AI offers token-based access to AI models, a low-cost way to run agent-style workflows — research, drafting, and summarising — without a $200-a-month plan. Together, the three tools cover the core Teammates loop: an agent to do the work, a data feed to keep it current, and affordable model access to power it.

The do-it-yourself route takes more setup than clicking "Create teammate," but for a freelancer or small team it delivers most of the value at a fraction of the cost.

Limitations to Know Before You Commit

Teammates are an interesting launch, but they are early. The feature is in research preview, so behaviour may change. It is Max-only, which prices out casual users, and web-only for now, with no iOS app yet. Teammates draw on Max's monthly credit allocation, but there is no published per-task cost, so heavy use could burn credits faster than expected. Human review is still needed for anything client-facing.

None of these is a reason to ignore Teammates — start with one small job, measure whether it saves you time, and only then expand.

Frequently Asked Questions

Is Perplexity Teammates free?

No. Teammates are available only to individual Max subscribers, at $200 per month (or $2,000 per year). There is no free tier, and the feature is in research preview.

Do I need the Max plan to use Perplexity Teammates?

Yes. As of launch, Teammates are limited to individual Max subscribers on the web; Enterprise Pro and Max plans are not included by default, and access is expected to expand.

What is the difference between Perplexity Teammates and Perplexity Computer?

Computer handles one-off, multi-step tasks in temporary sandbox sessions. Teammates are persistent: they keep working on an ongoing responsibility, remember goals, history, and shared files, and act when something changes.

Can Perplexity Teammates access my passwords?

Teammates can use saved credentials inside connected services, but they are encrypted and teammates cannot view them. You can review their work, take over the desktop, or redirect them mid-task.

When is Perplexity Teammates coming to iOS?

Perplexity has said iOS support is coming soon. At launch, Teammates are available only on the web for individual Max subscribers.

What AI model do Perplexity Teammates use?

The default model for a new teammate is Claude Opus 5.5, and you can change the model for each teammate or task.

Conclusion

The always-on agent era has arrived, and Perplexity Teammates are one of its clearest examples: agents that hold a responsibility, remember what they did, and keep working while you do something else. If you are on the Max plan, start with one small ongoing job — a morning inbox triage or a weekly competitor summary — keep a human in the review loop, and let feedback make it smarter each week.

And if the Max plan is more than your budget allows, you do not have to sit this shift out. An affordable workflow from tools on ProAccess.me — Manus AI for the agent work, Apify for live monitoring data, and MUSE AI for low-cost model access — gets you most of the same always-on value today. Pick one recurring job, automate it this week, and see how many hours it gives back.