Opening six tabs every morning to see what competitors shipped is a workflow problem. On November 6, 2026, Perplexity announced Teammates: agents you can give an ongoing responsibility instead of a one-off prompt. Its own example is watching competitors for pricing changes, launches, and hiring, then delivering a weekly summary. For a small founder team, the real question is what evidence that agent must collect before a signal reaches your roadmap.

What changed, and who gets access?

A teammate can plan work, use Computer tasks, files, connectors, Skills, and memory, and monitor connected tools for changes. The initial launch is a research preview for individual Max subscribers on the web. Free and Pro users cannot create one in this preview. Enterprise Pro and Enterprise Max are not included by default. Perplexity says the feature is still evolving. I have not tested it in a Max account; the access details and setup below come from its official documentation.

For an AI app builder, the useful shift is persistence. You can assign a teammate to follow Cursor, Lovable, and Replit for changes that affect how an internal app is built, connected, or published. A founder's post on X can suggest a lead, but the report should cite the original release note or documentation, plus its date and availability. That turns a stream of exciting posts into a small decision queue.

A 20-minute pilot

1. With an eligible individual Max account, open Perplexity on the web. Select the + next to Teammates in the sidebar and name a role such as “AI builder release radar.”
2. In Advanced, specify three products, your audience, a schedule, and a private output format. Connect only the tools needed for this pilot.
3. Give it two known examples: a real product release and a company update with no user-facing change. Check whether it correctly rejects the second.
4. Inspect every claimed release for source URL, date, rollout status, and a practical test. Missing fields should be marked unresolved.

Here is a starting instruction you can copy:

You are a private product radar for a founder building with AI.
On weekdays, monitor Cursor, Lovable, and Replit. Use X for discovery only.
Confirm each change in an official announcement, changelog, or documentation.
For each signal report: primary URL, date, availability (live/beta/gradual),
limits, affected users, and a 15-minute test I can run today.
Skip company news and internal changes without a reader decision.
Never claim to have tested a feature when you only read its docs.
Send the report to me PRIVATELY first. Do not publish, message anyone,
open a PR, or change an external tool without my approval.
If nothing useful changed, say "no actionable change".

That approval boundary matters. Perplexity documents that teammates can use connected tools and collaborate through Slack. For customer messages, publishing, or repository changes, define a human review step before connecting an account with write access. The report should distinguish confirmed facts from inferences and open questions.

Measure the pilot, not the number of links

For a week, count detected signals, signals backed by primary sources, false alarms, and decisions you actually changed. Ten links that change no experiment are noise. If an important release is missed, add its primary source and run the case again. Review usage credits as well: Perplexity says large projects, parallel agents, and frequent automations may use more of the monthly Max allocation. Access does not imply unlimited runs.

The win is a better prepared editor or builder, not an outsourced editorial judgment. The teammate gathers evidence; you decide what deserves a test and what deserves publication.

Sources and scope

Perplexity's November 6, 2026 changelog; official Teammates setup guide, last modified November 5, 2026. The pilot is a proposed editorial workflow, not a measured result from my own Max account.