![Perplexity Teammates: $200 Max Agents Launch [2026]](https://shattered.io/wp-content/uploads/2026/11/perplexity-teammates-ai-agents-ongoing-work-200-max-2026-gen.webp)
Perplexity put a new kind of worker into its product on November 4, 2026. Engadget reported on November 5 that Perplexity Teammates are the company’s AI agents for daily work tasks, and that the feature starts with Perplexity Max subscribers. The launch matters because it moves Perplexity away from the answer box and toward software that holds a job for days or weeks.
Perplexity’s own description is more precise than the headline. The company says Teammates are AI agents that own ongoing work, plan tasks, coordinate additional agents, and follow through until results are ready for a person to review. Examples in its materials include triaging bug reports, tracking metrics, researching a market, and managing growth campaigns. Access costs $200 per month or $2,000 per year on the Max plan.
That price and that persistence change the buying conversation. An agent that works in the background can also bill in the background, which is the worry behind our earlier look at the three agent limits Forrester wants. This analysis sorts what Perplexity and the press have confirmed from what remains rumor, then weighs the competitive and market consequences.
Don't miss new tech stories on Google
Add Shattered once in the Google app and our stories show up in your news suggestions.
What Perplexity Teammates Are and What Launched #
Perplexity Teammates are not a new chatbot skin. They are a delegation layer. A user hands a Teammate a responsibility, the Teammate breaks it into tasks, runs them, and comes back when the output is ready or when it needs a decision. The user reviews the result instead of steering every step.
Perplexity’s official announcement carries the date November 4, 2026. Other outlets covered the launch on November 5, which explains why you will see both dates in circulation. Neither date is wrong. One marks the company post and the other marks the news cycle that followed.
Availability starts narrow. Reports say the feature first reaches Perplexity Max subscribers on the web, and Engadget also reported availability on iOS. Perplexity Max was reported at $200 per month or $2,000 per year. Anyone on a cheaper tier would need to move up to try the feature, and the sources we checked do not describe a free trial.
The product sits on top of work Perplexity already sells. Its help documentation says each additional agent used for a task gets its own computer, which signals that Teammates borrow the company’s existing agent infrastructure rather than replace it. We unpack that design below.
Daily Work Tasks or Ongoing Work: Headline Versus Official Wording #
Engadget’s headline calls Teammates the company’s AI agents for daily work tasks. That is a fair shorthand for a news headline, but it is not Perplexity’s wording. The company describes Teammates as a team of AI agents that own ongoing work, and says they are personal agents that can take on long-running work.
The gap is more than semantics. “Daily” suggests chores that repeat every morning, such as summarizing an inbox. “Ongoing” and “long-running” suggest ownership of a project that stretches across many days, with state that persists between check-ins. The first is a scheduled prompt. The second is closer to a junior colleague with a standing brief.
We treat the daily-work framing as a characterization from the press, not a product claim. Nothing in the official materials we reviewed says Teammates are designed specifically for daily tasks. Readers comparing tools should hold Perplexity to its own wording, because the promise of long-running ownership is also a bigger promise to fail on.
It also changes how you should test the product. A one-day trial proves little about a tool built for work that unfolds over weeks. A fair evaluation hands a Teammate a real recurring responsibility and judges it on consistency, on how often it asks for input, and on whether its output survives review without heavy rework.
How Perplexity Teammates Plan, Delegate and Follow Through #
According to Perplexity, a Teammate does four things in sequence. It takes ownership of a stream of work, plans the tasks, coordinates additional agents when the job needs them, and follows through until results are ready for review. The coordination step is the part that separates this release from a single long chat.
Background work and check-ins
Teammates can run ongoing work in the background, monitor changes, and ask the user for input when needed. Those three behaviors define the experience. Background execution frees the user’s screen. Monitoring lets the agent react when something shifts, such as a metric moving or a new bug report landing. The request for input keeps a human in the loop at the points where judgment matters.
The design implies a notification stream. Someone has to decide how often a Teammate may interrupt, and what counts as important enough to interrupt. Perplexity’s materials, as reported, do not publish default thresholds, so early adopters will tune this by trial.
One computer per agent
The detail that stands out comes from Perplexity’s help documentation, which says each additional agent used for a task has its own computer. That means a Teammate that spins up helpers does not cram them into one shared session. Each helper gets an isolated working environment, which can limit how far one agent’s mistake spreads and can make parallel work faster.
It also raises a cost question. More computers mean more compute behind each project. Perplexity has not, in the sources we checked, explained how heavy usage is metered inside the Max price, so buyers should ask before they plan for dozens of concurrent agents.
Six Tasks Perplexity Names for Teammates #
Perplexity points to a specific set of jobs. Each one has a different review profile, and each one tests a different weakness of agent software.
- Triaging bug reports: The agent reads incoming issues, groups them, and flags the urgent ones. Review means spot-checking whether it merged two distinct bugs.
- Tracking metrics: The agent watches numbers and reports changes. Review means confirming it read the right dashboard and the right time window.
- Researching a market: The agent gathers sources and drafts findings. Review means verifying citations, a task Perplexity’s search heritage should help with.
- Investigating issues: The agent digs into a problem and reports a cause. Review means checking that the proposed cause fits the evidence.
- Testing product improvements: The agent tries changes and records results. Review means making sure the test was fair.
- Managing growth campaigns: The agent runs ongoing marketing work. Review means guarding the budget and brand voice.
The list skews toward work that produces a report or a recommendation, not work that moves money or ships code to production by itself. That is a sensible launch posture. A mistaken summary costs a reviewer a few minutes, while a mistaken ad spend decision costs real cash.
Long-form tasks also expose a known weakness in language models, which is drift over many turns. Our Claude versus ChatGPT drift test shows why a standing brief needs periodic re-anchoring, and a Teammate that runs for weeks will face the same pressure at larger scale.
Role Labels We Cannot Confirm #
Some third-party coverage describes Teammates under job-style labels, including Ads Manager, Engineer, QA Engineer, and Legal Research Assistant. We could not confirm those labels in the official Perplexity sources available to us. They may be examples, templates, or the outlet’s own shorthand. Until Perplexity documents them, treat them as unconfirmed.
The distinction affects expectations. A product with a fixed menu of professional roles implies pre-built workflows and perhaps guardrails per role. A product where the user defines the responsibility implies a blank slate and more setup. Which one Perplexity ships changes the learning curve and the risk profile.
| Claim | Status | Basis |
|---|---|---|
| Announcement dated November 4, 2026 | Confirmed | Perplexity’s official announcement |
| Launch covered November 5, 2026 | Confirmed as coverage date | Other outlets, including Engadget |
| Agents own ongoing work and follow through to review | Confirmed | Perplexity announcement and help documentation |
| Each additional agent has its own computer | Confirmed | Perplexity help documentation |
| Max at $200 per month or $2,000 per year | Reported | Engadget and other coverage |
| Web first, plus iOS | Reported | Initial coverage, with Engadget naming iOS |
| Built specifically for “daily work tasks” | Unconfirmed | Headline characterization only |
| Ads Manager, Engineer, QA Engineer, Legal Research Assistant roles | Unconfirmed | Third-party coverage, not confirmed officially |
| Team or enterprise plans | Not confirmed | Not described in sources we checked |
Perplexity Max Pricing and Availability #
TechRepublic’s coverage puts the entry cost at $200 for the Max tier, and Engadget gave both the monthly and annual figures. The annual option works out to $2,000, which is $400 less than twelve monthly payments of $200. That is a 16.7 percent discount for committing to the year, based on our arithmetic from the reported prices.
| Item | Figure | How we got it |
|---|---|---|
| Perplexity Max, monthly | $200 | Reported price |
| Perplexity Max, annual | $2,000 | Reported price |
| Twelve months billed monthly | $2,400 | $200 times 12 |
| Annual saving | $400, or 16.7% | $2,400 minus $2,000 |
| Effective monthly cost on annual plan | About $166.67 | $2,000 divided by 12 |
| Rough cost per working day | About $9.52 | $200 divided by an assumed 21 working days |
The per-day figure is our own estimate and rests on an assumed 21 working days. It frames the real question. If a Teammate saves a person ten minutes a day, the subscription pays for itself only at fairly high hourly rates. If it handles a task that would otherwise go undone, the math looks different. Value depends on the work you hand over, not on the sticker.
One more caution. Max is an individual plan in the reports we reviewed, so a ten-person team buying seats would multiply that $200 by ten. Anyone budgeting for a group should wait for clearer team pricing before committing.
What Perplexity and Its CEO Said #
The company’s launch language is consistent across channels. Perplexity’s announcement says: “Today, we’re launching teammates, personal agents that can take on that long-running work.” That line frames the product as a response to a specific pain, which is work that outlasts a single chat session.
The announcement also describes the intended division of labor. In Perplexity’s words, “Users can delegate ongoing actions to teammates so people can focus on outcomes rather than coordinating every task.” The promise is that humans set direction and judge results, while agents handle the coordination in between.
Perplexity’s launch post on X added operational detail: “Teammates work for you. They take on jobs, and work proactively using your Computer tasks, skills, memory, and tools. Teammates collaborate with other teammates and can message coworkers in Slack.” Two points stand out. The word proactively implies agents that start work without a fresh prompt, and the Slack mention implies that Teammates will reach into the places where teams already talk.
CEO Aravind Srinivas pushed the vision further. As The Verge reported, he said: “Over time, we envision every org having hundreds of teammates working for each person.” That is a statement of ambition, not a shipped capability, and it should be read as a long-range goal. It does reveal the commercial logic. If one person manages hundreds of agents, the subscription model has to stretch well beyond a single seat.
Competitive Comparison: Chatbots, One-Off Agents and Persistent Teammates #
Perplexity is not alone in chasing persistent agents. OpenAI, Anthropic, Google, and Microsoft all pitch agent products, and consumer-facing agent personas from Meta and OpenAI drew scrutiny in our piece on cute AI agents and their privacy risks. We do not have verified launch dates, pricing, or feature lists for rival persistent-agent products in the sources behind this story, so the table below compares categories of tool rather than named rivals.
| Category | Who starts the work | Typical duration | What the human does |
|---|---|---|---|
| Chat assistant | The user, every turn | One session | Prompts, reads, re-prompts |
| One-off agent task | The user, once per job | Minutes to hours | Launches the task, reviews the output |
| Persistent agent such as Perplexity Teammates | The agent, after a standing brief | Days to weeks, per Perplexity’s long-running framing | Sets the goal, answers questions, reviews results |
| In-editor coding agent | The developer, inside the IDE | Task by task | Accepts or rejects code changes |
The tricky competitor for Perplexity Teammates is not another startup. It is the assistant people already pay for. Our comparison of Perplexity, ChatGPT and Gemini for research shows how crowded the research-assistant field already is. Perplexity has to argue that a standing agent beats a very good chat window, and it has to argue it at $200 a month.
Perplexity’s edge, if it holds, comes from search. An agent that researches a market is only as good as its sourcing, and sourcing is the company’s origin story. The weak point is breadth. Rivals with deep ties to office suites, cloud platforms, or code repositories can plug agents into data that Perplexity must reach through connectors.
Historical Context: From Answer Engine to Agent Workforce #
Perplexity built its name as an answer engine that cites its sources. The pitch was simple. Ask a question, get a sourced reply, skip the ten blue links. Teammates extend that arc in a logical direction. After answers comes action, and after action comes ownership.
The wider industry has traveled the same road. Chatbots arrived first, then tool-using assistants, then agents that could browse and click, and now agents that persist between sessions. Each step handed the model more autonomy and handed the human a heavier review job. Our AI and machine learning coverage tracks how each lab frames that trade.
Persistence is the hard step, because it demands memory, scheduling, and trust. A chatbot that forgets you after the window closes does little damage. An agent that remembers, acts overnight, and messages your colleagues can do real damage if its instructions are wrong. That is why launches in this category tend to start as limited previews on premium plans, and Teammates follows that pattern.
The pricing choice fits that pattern. Perplexity placed its newest agent feature behind the $200 Max plan, which suggests the company expects Teammates to consume more compute than a typical chat subscriber does. That is our inference, since the company has not published its reasoning.
Market Impact: Seats, Compute and the Agent Race #
Digital Trends framed the launch with a warning in its headline, saying the new AI coworkers can take over your workload but do not come cheap. That tension defines the market impact. Agents promise labor savings, yet the cost of running them lands on a monthly invoice that rises with ambition.
For software vendors, a persistent agent threatens the per-seat model from two directions. If one person supervises many agents, fewer human seats may be needed. If the agents do the heavy lifting, vendors will want to charge for agent activity as well. We saw a version of that pricing debate in our comparison of [Windsurf and Cursor seat pricing](https://shattered.io/windsurf-vs-cursor-devin-desktop-rename-40-seat-2026/), where agent features pushed vendors to rethink what a seat means.
For Perplexity, the stakes are about revenue per user. A $200 plan only works if subscribers stay, and subscribers stay only if Teammates deliver results they would miss. The company has no proven retention data for this feature yet, and none appears in the sources we reviewed.
For buyers, the near-term effect is a new line item to justify. Finance teams that already scrutinize AI spend will ask what a Teammate produced last month, which means vendors need clear activity logs. The products that report outcomes in plain terms will win renewals over those that hide behind usage meters.
Risks: Permissions, Cost Control and Review Burden #
Three risks deserve attention before anyone hands a Teammate real responsibility. The first is access. An agent that monitors tools and acts on changes needs credentials, and every credential widens the attack surface. Our report on the Claude Dispatch flaw shows that agent products can ship with security gaps, so buyers should read the security documentation, not just the feature list.
The second risk is cost drift. Background agents run when nobody watches. Without hard caps, a poorly scoped brief could burn through resources while producing little value. Perplexity’s sources do not describe spending limits for Teammates, so users should treat the absence of published caps as an open question.
The third risk is review burden. A Teammate that produces ten outputs a day creates ten things to check. If checking takes as long as doing, the gain evaporates. The best deployments will pick tasks where verification is quick and errors are cheap, such as a weekly metrics digest, and avoid tasks where a wrong answer reaches customers unseen.
Privacy belongs on the list too. Agents with memory store context about people and projects. Anyone deploying Teammates inside a company should confirm where that memory lives and who can read it before connecting sensitive systems.
Who Should Try Perplexity Teammates Now #
The sensible early adopter is an individual who already pays for Perplexity Max and has a recurring research or monitoring chore. A founder tracking competitors, a product manager watching metrics, or a marketer reviewing campaign performance all fit. These users can judge output quality quickly and absorb a bad result without harm.
Teams should wait. The launch is aimed at individual Max subscribers, team pricing is not confirmed, and the governance features that a company needs, such as audit trails and permission scopes, are not described in the sources we checked. A pilot with one volunteer costs little, while a rollout across a department costs a lot.
A simple trial plan helps. Pick one responsibility with a clear definition of done. Run it for two weeks. Count the interruptions, the corrections, and the hours saved. If corrections outnumber saved hours, drop the task and try a narrower one. If results hold, expand slowly and keep a human reviewer on every output that leaves the building.
Five Predictions for Persistent AI Agents #
These are our judgments, not reported facts. Treat them as forecasts that could miss.
- Perplexity will widen access beyond Max. Research previews on premium tiers usually broaden once the compute cost per task drops, and a $200 gate limits the audience that can generate feedback.
- Rivals will ship persistent agents with office and code integrations. Vendors that own the documents, tickets, and repositories can offer deeper hooks than a search company can.
- Metering will get stricter. Flat-rate access to always-on agents strains margins, so expect usage caps or credit systems to appear within a few quarters.
- Audit logs and permission scopes will become the main buying criteria. Procurement teams will ask what an agent touched before they ask how smart it is.
- A public agent mishap will force a rules debate. As agents act on their own, a visible error involving money or customer data will draw regulator and press attention to the whole category.
The Takeaway on Perplexity Teammates #
Perplexity Teammates mark a real step from answers to ownership. The confirmed facts are modest and clear: agents that own ongoing work, plan and coordinate, run in the background, ask for input, and cost $200 a month on Max. The grander claims, from role labels to hundreds of agents per person, are either unconfirmed or still aspirations.
Readers should judge the product by what it proves over weeks, not by launch week headlines. If Teammates cut review time and hold quality, they will earn the price. If they create a pile of outputs that need checking, they will join the long list of agent demos that looked better than they worked.
Frequently Asked Questions #
What are Perplexity Teammates?
Perplexity describes Teammates as AI agents that own ongoing work, plan tasks, coordinate additional agents, and follow through until results are ready for review. Engadget summarized them as the company’s AI agents for daily work tasks.
When did Perplexity launch Teammates?
Perplexity’s official announcement is dated November 4, 2026. Other coverage reported the launch on November 5, 2026.
How much do Perplexity Teammates cost?
The feature was reported as initially available to Perplexity Max subscribers. Perplexity Max was reported at $200 per month or $2,000 per year. We found no separate Teammates fee in the sources we checked.
Where can I use Teammates?
Reports describe initial availability on the web, and Engadget also reported availability on iOS. Other platforms were not confirmed in our sources.
What tasks can a Teammate handle?
Perplexity’s examples include triaging bug reports, tracking metrics, researching a market, investigating issues, testing product improvements, and managing growth campaigns. Teammates can run work in the background, monitor changes, and ask the user for input.
Are roles like Ads Manager or QA Engineer official?
Not that we could confirm. Labels such as Ads Manager, Engineer, QA Engineer, and Legal Research Assistant appear in third-party coverage, but the official Perplexity sources we checked do not confirm them.
Do the extra agents share one computer?
No, according to Perplexity’s help documentation. Each additional agent used for a task has its own computer.
Is it worth paying $200 a month for Teammates?
That depends on the work you hand over. The feature is new, retention and reliability data are not published, and the value shows only if a Teammate cuts your review time on a recurring task. A two-week trial with one clear responsibility is the cheapest way to find out.