Perplexity Teammates are agents that keep working on a job
SummaryNov 4, 2026: Perplexity launched teammates, personal agents that own ongoing work, watch your connected tools and bring results back for review. Research preview on the web for individual Max subscribers ($200/month). Not tested by MODEL.
· MODEL Editors · 2 min read
Some links may be affiliate links. Disclosure

What launched. On Nov 4, Perplexity launched teammates: personal agents that take on a job instead of a single prompt. You give one a role, such as triaging bug reports, tracking metrics or researching a market. It plans the tasks, follows up as things change and brings finished work back for your review.
How it works. Teammates run inside Perplexity Computer and use your Computer projects, skills, memory and connected tools. Per Perplexity's help center, a teammate:
- breaks a goal into tasks and runs them in parallel across multiple agents, adding agents as the work grows
- gets its own isolated desktop computer and browser, so it can use sites without a connector and test a website end to end; you can take over and steer
- watches the tools you connect, such as Slack, GitHub and Linear, and starts work when something arrives (for example, answering review comments after it opens a pull request)
- keeps its role, schedules follow-ups and remembers between conversations
- lets you pick the model per teammate; new teammates default to Claude Opus 5.5
- can be tagged as an app in Slack, GitHub, Linear and more, and can hand work to other teammates in the same Project
Who it's for. People with recurring work they would otherwise prompt for every day: inbox and calendar triage, competitor tracking, support replies, bug triage that ends in a pull request.
Access and price. It is a research preview on the web for individual Perplexity Max subscribers. Perplexity's help center lists Max at $200/month or $2,000/year. Pro is not included, and Enterprise Pro and Enterprise Max are not included by default. Perplexity says iOS is coming soon.
The caveats. Max includes a monthly credit allocation, and Perplexity warns that complex projects use more credits, especially with parallel agents, large codebases or frequent automations. Perplexity also says the feature is still under development and behavior may change. Anything that watches your Slack, inbox and repos runs on the permissions you grant it: Perplexity says credentials are encrypted and teammates cannot view them, and recommends allowing only the GitHub repositories a teammate needs. Start narrow.
Early usage claim. Perplexity CEO Aravind Srinivas posted on Nov 7 that teammate automation runs grew 3.7x from launch day to day two. That is the company's own figure; MODEL has not seen the underlying data.
Disclosure: Not tested by MODEL. Not sponsored. No affiliate links.
MODEL Editors
Sources
- Teammates are agents for complex, ongoing work
Perplexity
- Get started with Perplexity teammates
Perplexity Help Center
- Teammates, fully local Computer on Mac, and Voice mode (changelog, 11/06/26)
Perplexity
- Perplexity Max
Perplexity Help Center
Keep reading

Builder tip
Builder tip: run Knip after your coding agent is done
We planted the leftovers coding agents tend to leave in a small TypeScript repo and ran Knip 6.44.2. It found the dead file, two unused packages, two unused exports and one import missing from package.json. Its auto-fix only does part of the cleanup.
· MODEL Editors

Tool Launch
OpenCode Browse lets your coding agent drive your real browser
Nov 6, 2026: OpenCode shipped Browse, an MCP server plus Chromium extension that lets agents use your real browser with your tabs, logins and extensions, and hands the tab back to you for logins, 2FA and payments. Not tested by MODEL.
· MODEL Editors

Tool Update
Qoduck: Qoder's free daily allowance on a stealth model for individual International users
Nov 7, 2026 (Singapore Time UTC+8): Qoder gives individual Qoder International users a daily free allowance on Qoduck — a stealth model whose identity is not disclosed. Within the allowance Credit rate drops 0.5x to 0.0x; Teams/Enterprise ineligible; desktop International only. Not tested by MODEL.
· MODEL Editors
FREE · EVERY WEEK
The top AI tools this week, tested.
Every week MODEL's editors pick the AI tools worth your time, try them hands-on, and tell you what's good, what's not, and who it's for. Free, in your inbox.
Free. Unsubscribe anytime. Our picks are editorial and can't be bought.