Signals · Capability Displacement
Perplexity's new teammates agents take over the follow-up work that fills a coordinator's week
Perplexity's teammates agents own ongoing follow-up, monitoring and chasing, so the human role that remains is to specify the work and sign it off.
by Jo·3 min read·
New here? Start with the free AI Survival Kit →
Perplexity's new teammates agents take over the follow-up work that fills a coordinator's week
0:00 / 4:52
This voice is generated by AI.
Part of the guide Will AI take your job? How to find out for your role, in an afternoon
Perplexity announced this week on its company blog that it is launching "teammates", personal agents built to own long-running work instead of answering one query at a time. A teammate plans and coordinates projects. It remembers goals, task history and shared files across runs, monitors work without waiting for a fresh prompt, and brings completed work back for review. Perplexity names marketing, sales, support, operations and finance as the workstreams. Teammates launch first on the web for Perplexity Max subscribers, with iOS to follow.
What the launch actually targets
The headline reads as a product launch. The detail reads as a job description.
Perplexity's own examples make the point. The sales teammate notices when a regular customer stops transacting, reviews their order history and support issues, then drafts an offer or flags a problem to fix before anyone reaches out. The operations teammate chases missing items when a deadline is 72 hours away. The support teammate reviews the previous day's complaints, summarises trends and flags the cases that need a personal response from the CEO.
These are follow-up, chasing and monitoring tasks. Together they make up the coordination layer that fills an account manager's or operations assistant's week. Perplexity says it directly: users delegate ongoing actions "so people can focus on outcomes rather than coordinating every task."
This is how a role gets hollowed out. The account coordinator keeps the title after this launch. The tasks that justified the headcount start moving to a system that remembers the agency needs two weeks' notice and builds that lead time into every schedule after it.
Where the human stays in the loop
The same post shows what Perplexity leaves with people. Every example ends at a review point. Teammates "present results for review" and share "progress updates and approval requests". Users can "take over the desktop, or redirect a teammate as it works." The person setting up a teammate has to describe the work it should own: where the work comes from, what it should deliver and when it should act.
The work that remains is specification, judgement and sign-off. Someone has to define what counts as a regular customer going quiet. Someone decides whether the offer goes out, and owns the result when it fails. That is the layer above task execution, and accountability sits there. AI does the routine work. You do the thinking.
The weak spot for a professional is the middle. If you still do the follow-up by hand while a teammate could do it, and you have no authority to approve what it produces, you hold the part of the job that is moving.
Your Next Move
- Audit your coordination load this week. List every recurring follow-up, reminder, status chase and monitoring check from your last five working days. Those are the tasks in Perplexity's examples. Our method for working out which of your tasks AI absorbs first, and scoring your role turns that list into a result you can act on.
- Write one brief as if you were setting up a teammate. Perplexity's setup asks where the work comes from, what it should deliver and when it should act. Write that brief for one of your workstreams. If you cannot write it, you have found judgement work you have not yet made explicit. Build there first.
- Move toward the approval step. Ask to own the sign-off on one process, such as the outreach that goes out, the vendor escalation or the list of complaints for the CEO. Whoever reviews the agent's work holds a position the agent cannot fill.
I build systems like the one publishing this site. → Work with me
About the author
Jo
Jo runs The War Room: one signal a day on how AI is changing work, and what to do about it.
Sources
Get the Briefing
Want more intelligence like this?
The weekly briefing, free — and the AI Survival Kit with it.
More in Signals
See allSignals · Capital
US tech giants have $3.2tn in hidden AI obligations, and someone has to make them pay back
4 min read
Signals · Agent Access
Delta, Chase and Amazon are shutting out AI agents, and every task you delegate now depends on the other side's permission
4 min read
Signals · Accountability
Google DeepMind resignations show where human oversight of AI breaks down
4 min read