Most AI tools wait for you to ask. You type a prompt, you get an answer, and the conversation ends. Perplexity Teammates are built on the opposite idea: you hand an AI agent a job, and it keeps doing that job — day after day, week after week — until you tell it to stop. It watches your inbox, tracks your competitors, drafts your support replies, and messages your colleagues in Slack when it needs a decision. You do not prompt it. You manage it.
Perplexity launched Teammates on 4 November 2026, with chief executive Aravind Srinivas announcing the wider rollout the following day. The feature is a research preview, available only to individual subscribers on the $200-a-month Max plan, and only on the web for now. That price alone makes it the most expensive way to hire an AI helper that any major AI company currently sells — and it raises the obvious question: what exactly are you getting for $200 a month, and is it worth it?
This guide explains what Perplexity Teammates are, how they work inside Perplexity Computer, what they can and cannot do today, what the real cost looks like once the credit meter starts running, and how they compare with the always-on agents from OpenAI, Anthropic and Salesforce. Every figure about pricing and credits comes from Perplexity’s own published pages and launch reporting; where a number is a company claim rather than an independently tested result, this article says so.

Table of Contents
- What Are Perplexity Teammates?
- How Perplexity Teammates Work
- What Can a Perplexity Teammate Actually Do?
- Perplexity Teammates vs Perplexity Computer
- Perplexity Teammates Pricing: The $200 Question
- The Credit Math: What One Month of Teammates Really Costs
- Perplexity Teammates vs the Competition
- Who Should (and Should Not) Pay $200 a Month
- How to Set Up Your First Perplexity Teammate
- Limitations and Risks to Know First
- Is Perplexity Teammates Worth It?
- Perplexity Teammates: Frequently Asked Questions
- The Bottom Line on Perplexity Teammates
What Are Perplexity Teammates?
Perplexity Teammates are persistent AI agents that take ownership of an ongoing piece of work. Instead of answering one question and stopping, a teammate holds a standing role — “triage the support inbox”, “track what our competitors ship”, “watch this deal and keep the file current” — and keeps working on it between your conversations. It remembers what it did last week, what it tried, which files it used, and what it is still waiting on. When something changes, it reacts. When it needs a human decision, it asks.
Perplexity describes the shift in one sentence: you stop writing prompts in a session and start giving a goal instead. That is the difference between a chatbot and a teammate. A chatbot is a tool you pick up. A teammate is a colleague you brief.
Each teammate runs inside Perplexity Computer, the company’s agent system that has been assigning multi-step tasks to AI agents since February 2026. Computer gives a teammate its own isolated desktop and browser, its own saved credentials, and access to the projects, skills and memory you have already built in Perplexity. The teammate inherits that context, so it does not start from zero every morning. It also keeps the role you gave it until you change it — the assignment is the persistent part, and that persistence is the whole product.
Perplexity Teammates Are Not Chatbots With a Memory Toggle
It is tempting to describe Perplexity Teammates as a chatbot that remembers things. That undersells them in one important way and oversells them in another. What makes a teammate different is not memory alone — it is initiative. A chatbot with memory still waits for you. A teammate works while you are doing something else, checks its own results, and comes back to you when there is something to review or decide.
The oversell is the word “teammate” itself. These agents do not have judgement, taste, or accountability in the human sense. Perplexity is clear that finished work still goes to a person for review, and the help centre warns that the feature is still under development and that its behaviour may change. A teammate is a capable, tireless junior worker with no common sense — which is exactly why the review step matters and why the setup section of this guide spends time on boundaries. If you are new to AI assistants generally, our best AI chatbot guide explains how the one-off chat tools differ from the agents covered here.
Where Perplexity Teammates Sit in the Always-On Agent Race
Perplexity is not first to this idea, and it is not alone. OpenAI has been building always-on agents under its Dots programme, Anthropic has Claude Duet, and Salesforce has pushed its Agentforce Coworker into Microsoft Teams. The race is the same for all of them: whoever builds the most trustworthy always-on worker wins the most valuable seat in the workplace — the seat that does the recurring work nobody wants to do twice.
What makes Perplexity Teammates distinctive in that race is the starting point. Perplexity began as a search and research company, so its agents start from research strength: finding information, checking sources, and synthesising what changed. That is a natural fit for the jobs Perplexity pitches first — competitor tracking, market research, and inbox triage — and a weaker fit, for now, for jobs that need deep access to a company’s own operational systems. The comparison section later in this guide puts the rivals side by side.
How Perplexity Teammates Work
A teammate’s working day has a shape, and understanding that shape is the fastest way to understand the product. The cycle has four stages: plan, execute, inspect, and hand back. Perplexity’s help centre describes the same loop, and it is worth learning because it explains both what teammates are good at and where they slow down.
First, you give the teammate a goal and a role. You name it, describe the work in plain language, connect the tools it needs — Slack, GitHub, Gmail and others — and assign it to a project. The project is where the shared context lives: the files, the skills, and the memory the teammate draws on. A teammate without a project is a worker without a desk.
Second, the teammate breaks the goal into tasks. It works out which tasks can run at the same time and starts extra agents for those tasks. Each added agent gets its own computer and works on its own task from start to finish. A large job — say, researching five competitors and drafting a comparison — can therefore run in parallel instead of in one long queue. This is where the credit meter starts to matter, because parallel agents cost more than one agent working alone, and the pricing section below works through the numbers.
Third, the teammate inspects its own results. It checks what the sub-agents produced, decides whether to add more agents or change the plan, and adjusts when priorities shift. This is the part that separates Perplexity Teammates from a simple scheduled task. A scheduled task runs the same instructions every time. A teammate is supposed to notice when the instructions no longer fit and adapt.
Fourth, the finished work comes back to you for review. Nothing is published, sent, or filed without a human looking at it first — at least, nothing should be. The teammate can draft a support reply, prepare a fix, or propose a response to a stalled deal, but the human approves it. That review step is both a safety feature and a speed limit: it keeps mistakes from going out under your name, and it means a teammate never truly works unsupervised.
Perplexity Teammates and Slack: Messaging Your Colleagues
The most talked-about capability of Perplexity Teammates is the Slack connection. A teammate can be added to a Slack channel, tagged like an app, and given the ability to message coworkers directly. When it detects something worth flagging — a deal that has stalled, a bug report that matches a known issue, a competitor announcement that changes your roadmap — it can ask a human for a decision in the channel where that human already works.
Teammates can also pass work to each other. One agent finds bugs on a website and hands the findings to another agent that prepares the fixes. That handoff is the closest thing in the product to a real team: specialised workers, each with a role, passing work along instead of one generalist doing everything badly.
There is a social question here that no launch post answers. A Slack message from an AI agent looks like a message from a colleague, and coworkers will need to know which is which. Teams that adopt Perplexity Teammates should agree, before the first message goes out, how a teammate identifies itself in shared channels — because a silent AI participant in a human conversation is a trust problem waiting to happen.
Which Model Powers a Perplexity Teammate?
New teammates start on Claude Opus 5.5, Anthropic’s flagship model, and you can choose a different model per teammate or per task. That choice matters more than it looks. Opus is a strong reasoning model, which suits the planning and inspection stages of the teammate loop, but it is also an expensive model to run — and the credit meter, not the model name, is what decides your monthly bill. Perplexity has not published a per-model credit rate for teammates, so the practical advice is simple: start with the default, measure what a week of real work costs, and only then experiment with cheaper models for routine jobs.
What Can a Perplexity Teammate Actually Do?
Perplexity pitches Perplexity Teammates at recurring jobs with a clear source, a clear output, and a rhythm. The launch examples are concrete: triaging an inbox, tracking competitors, and answering support messages. But the useful question is not what the marketing page lists — it is which jobs survive contact with a real week of work. Here is how the main use cases actually play out.
Inbox Triage: The First Job Most People Will Try
Inbox triage is the safest first job for a teammate. The inputs are clear (incoming mail), the outputs are clear (sorted, labelled, drafted replies), and the cost of a mistake is low because nothing is sent without review. A teammate connected to Gmail can read new messages, sort them by urgency, draft replies to the routine ones, and flag the ones that need a human — a refund request with an angry customer, a contract question, a message from a key client. The human still sends the replies. The teammate does the sorting and the first draft, which is where most of the time goes.

Competitor Tracking: The Job Perplexity Is Best Built For
Competitor tracking is the job Perplexity Teammates are most naturally good at, because it plays to Perplexity’s core strength: research. Give a teammate a list of competitors and a rhythm — daily, or every Monday — and it watches for product launches, pricing changes, press coverage, and hiring signals, then summarises what changed and why it matters. The output is a briefing, not a raw feed, and the briefing is where the value sits. Anyone can set up a Google Alert. A teammate is supposed to tell you which alert actually matters.
Support Replies: Fast Drafts, Human Approval
Support work is the highest-volume use case Perplexity names, and the one that needs the most careful boundaries. A teammate can read incoming support messages, look up past exchanges and known issues, and draft a reply. For a small business drowning in “where is my order” messages, that draft alone is worth real money. But support replies carry reputation risk: a wrong answer sent under your company’s name is worse than a slow answer. The sensible setup is a narrow one — let the teammate draft, let it suggest, and keep the send button firmly in human hands until the drafts have earned trust over several weeks.
Bug Triage and Code Tasks: Useful, With a Watchful Eye
For software teams, a teammate connected to GitHub can watch new issues, reproduce the simple ones, label them, and group duplicates. It can also hand findings to a second teammate that prepares fixes. This is the multi-agent handoff Perplexity demonstrates, and it is genuinely useful — but it is also the use case where credit burn rises fastest. Perplexity’s own help centre warns that consumption climbs with parallel agents, large codebases, and frequent automations. A bug-triage teammate that spins up five agents on a large repository can burn through a day’s credit budget before lunch. Start with one repository, one label set, and a spending cap.
Market Research and Deal Watching
The deal-watching example in Perplexity’s launch material is worth noting because it shows the teammate loop at its best. When a teammate detects that a deal has stalled, it looks into past exchanges and support issues, drafts proposals and possible responses, and asks a human for approval. That is not a search query — it is a small investigation, triggered by a change, ending in a decision point. Market research works the same way: a standing brief, updated as the market moves, rather than a one-off report that goes stale the day after it is written.
Perplexity Teammates vs Perplexity Computer
Confusing Perplexity Teammates with Perplexity Computer is easy, because teammates are built on Computer and use the same projects, skills and memory. The difference is the unit of work. Computer is the agent that does a task. A teammate is the agent that owns a job.
With Computer, you ask for something — research this, build that, book this — and the agent works through it, possibly across several models, and returns a result. The job ends when the result is delivered. With Perplexity Teammates, you assign a role and the agent keeps it. The work does not end; it recurs. The teammate checks for changes, starts new tasks when the situation changes, and carries the history forward. Computer is a contractor you hire per job. A teammate is a member of staff with a job description.
The practical difference shows up in what you manage. With Computer, you manage prompts. With teammates, you manage roles, tools, and boundaries: which Slack channels it can post in, which GitHub repositories it can read, which files it can touch, and what it must never do without approval. That is a different kind of management, and it is closer to managing a person than to writing a good prompt.
One more difference matters for buyers. Computer is available on the Pro plan at $20 a month, and since the November 2026 update it can run fully on supported Apple-silicon Macs for Pro, Max and Enterprise users. Perplexity Teammates are not on Pro at all. They are Max-only, web-only, and in research preview. If you already pay for Pro and like Computer, teammates are not an upgrade to your current plan — they are a tenfold price jump to a different one.
Perplexity Teammates Pricing: The $200 Question
There is no separate price for Perplexity Teammates. The price is the Max plan, and the price is $200 a month on monthly billing, or $2,000 a year — about $167 a month — on annual billing. Pro subscribers must upgrade. Enterprise Pro and Enterprise Max accounts do not get teammates by default; an organisation’s administrator has to request access. The research preview is for individual Max subscribers on the web, with iOS coming soon and native mobile and desktop apps promised later.
But $200 is only the entry ticket. The real cost of Perplexity Teammates is measured in credits, and the credit system is where most buyers will be surprised — in both directions.
How Perplexity Credits Work
Max includes 10,000 Computer credits a month, plus a one-time bonus of 35,000 credits when you sign up, which expires after about 30 days. One hundred credits equal one dollar, so the monthly allowance is worth $100 of agent work inside a $200 subscription. Unused monthly credits expire on your billing date. The default spending cap is $200 a month and can be raised to $5,000. Auto-refill is off by default — if you want it, you must set it explicitly, choosing both the refill amount and a monthly cap.
Teammates draw on that same monthly credit pool. At launch, Perplexity is offering a promotional period of free teammate usage for Max subscribers, after which teammates consume credits like any other Computer work. Perplexity has not published how long the promotional period lasts or how large the grant is. That is the single most important unknown in the whole pricing picture: the cost of running a teammate today is promotional, and the cost in a few months will be metered. Anyone budgeting for teammates should budget for the metered price, not the promotional one.
When credits run out, running tasks pause and new tasks do not start. That is a hard stop, not a gentle slowdown — which is actually a good design choice. A teammate that silently keeps working on a credit card you forgot about would be worse than one that stops and tells you.
The Credit Math: What One Month of Teammates Really Costs
Perplexity publishes reference values for what different kinds of Computer work cost in credits. Those figures are for Computer tasks, not teammates specifically — Perplexity has published no typical cost for one teammate doing one real job for one month — but they are the best available guide to what the meter will do once the promotional period ends.
| Type of task (Perplexity’s reference values) | Credits | Dollars |
|---|---|---|
| Quick fact or definition | 15–70 | $0.15–$0.70 |
| Multi-step research report | 350–950 | $3.50–$9.50 |
| Data dashboard | 2,400–9,800 | $24–$98 |
| Entire app or multi-day build | 15,000+ | $150+ |
Now translate that into a teammate’s month. Your 10,000 monthly credits cover roughly 10 to 28 research reports, depending on complexity — or a single data dashboard and a handful of reports, or about two-thirds of one multi-day build. A teammate that triages an inbox every morning, checks competitors daily, and drafts a weekly briefing will use credits every single day, and parallel agents multiply the burn. One independent analysis of the launch noted that a teammate running the wrong task in the wrong way could burn through the whole 10,000-credit monthly allowance in under an hour.
That is an extreme case, not a typical one — but it shows why the spending cap exists and why you should set it before the first teammate starts work.
The honest summary is this: the $200 subscription buys the seat. The credits buy the work. A light user who runs one teammate on one narrow job will likely stay inside the allowance. A heavy user who gives three teammates broad jobs with parallel agents will pay more — potentially much more — and should expect to. Perplexity’s own framing is that $100 of your $200 is agent work, and everything above that you buy at the same rate. Budget accordingly, and measure the first month before you commit to the second.

Perplexity Teammates vs the Competition
Perplexity Teammates do not exist in a vacuum. The always-on agent is the most contested idea in AI software right now, and every major lab has an answer. The differences between them are less about the idea — all of them promise an agent that keeps working — and more about the starting point, the price, and who is allowed in.
Perplexity Teammates vs OpenAI Dots
OpenAI’s Dots programme is the closest rival in spirit: persistent agents that own ongoing work rather than answering single prompts. OpenAI’s advantage is distribution — ChatGPT is where most people already work — and its agent can reach deep into the tools people already use. Perplexity’s advantage is research depth and the credit-transparent pricing model. The honest difference for a buyer is this: if your work lives in ChatGPT, Dots is the natural choice; if your work starts with research and you want to see the meter running, Perplexity Teammates are the more transparent option. Neither product has published independent benchmarks for always-on work, so neither can prove it is better — only different.
Perplexity Teammates vs Anthropic Claude Duet
Anthropic’s Claude Duet is the other named rival, and it has one structural advantage: it is built by the company that also builds Claude Opus 5.5, the model Perplexity Teammates run on by default. That does not make Duet better by definition, but it means Anthropic controls the whole stack — model, agent, and interface — while Perplexity is building an agent layer on top of a model it does not own. Perplexity’s counter-argument is flexibility: you can change the model per teammate or per task, and you are not locked into one lab’s model.
For a business audience, the more practical distinction is availability. Both products launched in research preview in the same season; both are gated behind premium plans; both are still proving themselves. The site’s earlier coverage of Claude Personal Finance shows the same pattern at Anthropic: powerful capability, careful rollout, premium price.
Perplexity Teammates vs Salesforce Agentforce Coworker
Salesforce’s Agentforce Coworker is the enterprise answer to the same question, and it now works inside Microsoft Teams — where most large companies actually spend their day. Agentforce is built for the enterprise stack: CRM data, service cases, sales pipelines, and the approval chains that large companies require. Perplexity Teammates are built for the individual professional and the small team: faster to set up, cheaper to start, and far less entangled with enterprise procurement.
The comparison is really a question about who you are. If you need agents inside Salesforce and Teams with enterprise audit trails, Agentforce is the product. If you need one agent to watch your inbox and your competitors, Perplexity Teammates are the simpler tool — and the cheaper one, at $200 a month against Agentforce’s per-user enterprise pricing.
One gap is worth naming directly. Independent reporting on the launch noted that the audit trail — the record of what an agent did — is gated behind 50 seats or one Enterprise Max seat. That matters. An agent that acts in Slack and GitHub without a readable log is an agent you cannot fully supervise, and supervision is the whole job of managing a teammate. Enterprise buyers should treat that as a blocking question, not a footnote.
Who Should (and Should Not) Pay $200 a Month
The price draws a hard line, and it is worth being blunt about who is on which side of it.
Perplexity Teammates are a good fit if: you already pay for Max and already run Computer daily; you have one or two recurring jobs with a clear source and a clear output; you are a founder, a researcher, or a small-team lead who spends hours a week on inbox triage, competitor watching, or support drafting; and you are willing to spend the first month measuring credit use before you scale up. For that person, the promotional period is a genuine free trial of a genuinely new way of working.
Perplexity Teammates are a poor fit if: you are on the $20 Pro plan and would be upgrading tenfold for one feature; you need enterprise access, audit logs, or admin controls today; your work needs agents inside systems Perplexity does not yet connect to; or you want an agent that works without a human review step. They are also a poor fit if the job you have in mind is a one-off project — that is Computer’s job, at a tenth of the price.
The most common mistake a buyer could make is upgrading to Max for teammates alone, without already being a heavy Computer user. If you do not already burn through Computer credits every month, you are paying $200 for a feature whose meter you have not yet learned to read. Try Computer on Pro first. If you outgrow it, the upgrade will feel obvious instead of speculative.
How to Set Up Your First Perplexity Teammate
Setting up a teammate takes three steps, according to Perplexity — and then a fourth step that Perplexity does not emphasise enough: setting the boundaries.
Step 1: Connect the tools. Link the plugins the teammate needs — Slack, GitHub, Gmail, and any others it will read from or write to. Connect only what the job requires. A competitor-tracking teammate does not need your Gmail. A support-drafting teammate does not need your code repositories. Every connection is a door, and a teammate should only have the keys it needs.
Step 2: Define the role. Give the teammate a name, describe the work in plain language, and assign it to a project where the shared files, skills and memory live. Be specific about the output — “a 300-word briefing every Monday at 9 a.m.” beats “keep me updated on competitors” — and be specific about what it must never do: never send a message without approval, never touch a production system, never share a file outside the project.
Step 3: Start small and watch the meter. Give the teammate one narrow job for the first week. Watch how many credits it uses, read everything it produces, and check the review queue daily. Perplexity’s own guidance is to measure before scaling, and the credit reference values in this article give you a yardstick. If a week’s inbox triage costs 2,000 credits, you know what a month costs. If it costs 9,000, you know the job is too broad.
Step 4: Set the spending cap before you need it. The default cap is $200 a month, and auto-refill is off by default. Keep it that way until you have a month of real data. A cap is not a limitation on the product — it is the difference between a teammate and a runaway bill.

Limitations and Risks to Know First
Perplexity Teammates are a research preview, and that label is doing real work. Here is what it means in practice.
It is web-only, for now. Teammates work on the Perplexity website at launch, with iOS coming soon and no date attached. If your team works on mobile, the teammate is not there yet.
It is Max-only, and Enterprise is not included by default. Pro subscribers are excluded, and enterprise accounts need an administrator to request access. That is a deliberate gate, not an oversight — but it means most teams cannot even try the product today without a $200 individual subscription per user.
The promotional pricing will end. Teammates are free for Max subscribers during a promotional period whose length and size Perplexity has not published. When it ends, teammates draw on the 10,000 monthly credits. Nobody outside Perplexity knows what a continuously running teammate costs yet, because nobody has run one for a full metered month. Treat every cost claim — including the ones in this article — as an estimate until the first metered bills arrive.
There are no benchmarks and no independent tests. Perplexity has published no benchmarks for Perplexity Teammates, which is understandable for a product launch but means the honest answer to “how good is it?” is that nobody outside Perplexity knows yet. The internal figures Perplexity’s chief executive shared — about 10,000 teammates built inside the company in 30 days, a median employee getting seven times more agent-hours a week than in September, one in five reaching 40 or more agent-hours a week — are the company’s own claims, not independently verified results. They show enthusiasm inside Perplexity. They do not prove the product works for your team.
The review step is a feature, not a bug — but it is also a bottleneck. Every finished job comes back to a human. That keeps quality up and prevents mistakes from going out under your name. It also means a teammate never truly works while you sleep, because the work waits for your review in the morning. The “always-on” promise is real for the doing, and limited for the finishing.
Behaviour may change. The help centre warns plainly that the feature is still under development. Workflows you build this month may behave differently next month. Do not build a business-critical process on a research preview without a fallback.
Is Perplexity Teammates Worth It?
The answer depends on what you compare it with. Compared with hiring a person to do the same recurring work, $200 a month is absurdly cheap — a part-time assistant costs ten times that, and does not work at 3 a.m. Compared with doing the work yourself, $200 a month is expensive — unless the hours it saves are worth more than $200 to you, which, for most professionals who bill their time, they are.
The comparison that actually matters is with the alternatives. A Pro subscriber at $20 a month with Computer can already do most one-off jobs. The $180 difference buys persistence: the standing role, the memory between runs, the watching, and the Slack messages. That persistence is worth $180 a month only if you have recurring work that genuinely recurs — daily triage, weekly briefings, continuous watching. If your work is mostly one-off tasks, keep the $20 plan and spend the difference on something else.
There is also a timing argument. Teammates are a week old, in research preview, with promotional pricing and unpublished metered costs. The people who benefit most from trying it now are the people who can afford to measure it — Max subscribers who already run Computer daily and have a narrow job to hand over. Everyone else can afford to wait: the product will still be there in three months, the pricing will be clearer, the first metered bills will have been published, and the rough edges will be smaller. Waiting is not missing out. It is letting someone else pay the early-adopter tax.
Perplexity Teammates: Frequently Asked Questions
What are Perplexity Teammates?
Perplexity Teammates are always-on AI agents that own an ongoing piece of work — such as inbox triage, competitor tracking, or support replies — instead of answering one prompt at a time. They run inside Perplexity Computer, keep their role and memory between runs, and hand finished work back to a human for review.
How much do Perplexity Teammates cost?
Perplexity Teammates require the Perplexity Max plan: $200 a month on monthly billing or $2,000 a year (about $167 a month) on annual billing. Max includes 10,000 Computer credits a month, and teammates draw on those credits after a promotional free-use period whose length Perplexity has not published. One hundred credits equal $1.
Are Perplexity Teammates available on Pro?
No. Perplexity Teammates are available only to individual Max subscribers, on the web, as a research preview. Pro subscribers must upgrade to Max, and Enterprise Pro and Enterprise Max accounts do not get the feature by default — an administrator has to request access.
Which AI model powers Perplexity Teammates?
New teammates start on Claude Opus 5.5 by default. You can choose a different model per teammate or per task.
Can Perplexity Teammates message coworkers in Slack?
Yes. A teammate can be added to a Slack channel, tagged like an app, and given the ability to message coworkers directly — for example, to ask for a decision when a deal stalls. Teammates can also pass work to each other.
Do Perplexity Teammates work on mobile?
Not yet. Teammates are web-only at launch, with iOS coming soon and native mobile and desktop apps promised later. Perplexity has not given dates.
Is Perplexity Teammates worth $200 a month?
For a Max subscriber who already runs Computer daily and has a narrow recurring job to hand over, the promotional period makes it worth trying. For a Pro subscriber upgrading tenfold for one feature, or a team that needs enterprise audit logs today, the answer is to wait — the pricing will be clearer and the product more mature in a few months.

The Bottom Line on Perplexity Teammates
Perplexity Teammates are the clearest expression yet of where AI work tools are heading: away from the prompt box and toward the job description. The idea is sound, the launch is honest about its limits, and the research strength behind it is real. A teammate that watches your competitors, triages your inbox, and drafts your replies — and remembers what it did last week — is a genuinely different kind of tool from a chatbot that waits to be asked.
But the product is a week old, locked behind the most expensive plan Perplexity sells, running on promotional pricing with an unpublished meter, and without a single independent test behind it. The $200 subscription is only the entry ticket; the credits are the real bill, and nobody outside Perplexity knows what a month of real teammate work costs yet.
That is not a reason to dismiss Perplexity Teammates. It is a reason to try them the way Perplexity itself suggests: one narrow job, one week, the meter watched closely — and to keep the send button in human hands until the teammate has earned it. The always-on agent race has just added its most transparent competitor. Whether it is worth $200 a month is a question only your first metered bill can answer.
