What Are Asana AI Teammates? Meeting Prep and Follow-Up Tasks

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for What Are Asana AI Teammates? Meeting Prep and Follow-Up Tasks.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for What Are Asana AI Teammates? Meeting Prep and Follow-Up Tasks.

Asana AI Teammates are Asana's prebuilt AI agents (30-plus ready-made roles, or your own) that you assign tasks to like a colleague, on any paid plan. For meetings, give one a recurring prep task the day before, then hand it the notes afterwards to create owned follow-up tasks and nudge overdue ones. Each successful run uses one AI request.

The Complete Excel Course with AI Integration

From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.

By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner

Buy now

Requests are what limit a small team, not features. Starter and Advanced include 5 requests per user per month, capped at 50 per account, and that pool is shared with Asana Dash. Extra requests cost $0.50 each prepaid or $0.60 on pay as you go. A daily chase run alone would use around 20 a month, so plan the schedule on paper before you switch anything on.

What is an Asana AI Teammate, and how does it differ from Dash, AI Studio and AI Meetings?

An Asana AI Teammate is a shared agent set up for a role: it gets assigned tasks, reads and writes comments, and shows up in the same activity feeds as people. Asana now has four AI features that overlap, and picking the wrong one is the usual way to waste money. Asana's engineering write-up explains that when a teammate is assigned a task it receives the task's context (parent project, related goals, dependencies, collaborators) and can search wider across the work it has been given access to.

FeatureWhat it's forWho uses itWhat it costsPlan
AI TeammateA role-based agent that takes on tasks, comments and creates workShared by the teamOne AI request per successful runStarter and above
Asana DashYour personal assistant: answers questions, drafts, hands work to a teammateOnly youA request only for complex work such as long-form writing, web search or a hand-offStarter and above
AI StudioNo-code automations for structured steps like routing and checksRuns in the backgroundAI Studio credits, a separate poolStarter and above
AI MeetingsTranscribes calls, writes summaries, suggests tasksMeeting ownerOne request per transcript up to 60 minutesAdvanced and above

For meeting prep and chasing, a teammate is the right tool because both jobs repeat every week and need judgement over project data: which renewal matters most, which follow-up is really late. A fixed reminder ("due tomorrow, notify the assignee") is a free rule, not a teammate; the five starter automation rules cover that. And a one-off "what happened in last month's meetings?" is a Dash question.

How many AI requests does a six-person team get on Asana Starter?

A six-person team on Asana Starter gets 6 x 5 = 30 AI requests a month, because the 50 cap doesn't bite until ten users. Take a six-person insurance broker on Starter. On annual billing that's 6 x $10.99 = $65.94 a month. The team holds one meeting a week that matters: Thursday's renewals review, covering every policy renewing in the next 60 days.

Here is the first plan someone sketched, and why it didn't survive the sum:

JobRuns per monthRequests
Prep the Thursday agenda (Wednesday afternoon)4 to 54 to 5
Turn Thursday's notes into follow-up tasks4 to 54 to 5
Chase overdue follow-ups every weekdayabout 22about 22
Total30 to 32

That plan uses the whole allowance and leaves nothing for anyone's own use of Dash. Moving the chase to Monday, Wednesday and Friday cuts it to about 13 runs, so the monthly total falls to 21 to 23, with seven or more spare. If the team later decides the daily chase is worth it, usage goes back to 30 to 32 a month: at most two requests over the allowance, or $1.20 on pay as you go, but with nothing left for anyone's own Dash work. Small money either way, but it's better decided than discovered.

Two limits to know before relying on the schedule. Asana emails admins at 65%, 85% and 100% of the allowance, and those thresholds can't be changed. At 100%, request-based actions pause until the reset or until extra usage is switched on. Asana's community team said that a recurring task assigned to a teammate showing "paused" is usually exactly that.

If you want to model the monthly request budget in a spreadsheet, try The Complete Excel Course with AI Integration on Coding Liquids.

How do you set up an Asana AI Teammate for meetings?

Setting up an Asana AI Teammate for meetings takes about an hour, and none of it works until AI is on for your domain. Only a super admin can switch Asana AI on; regular admins can turn AI features off but not back on. Depending on your plan, a user may also need an AI Teammates licence assigned by an admin. Then:

  1. Pick a starting teammate. Open the Agents tab in your renewals project, where Asana suggests teammates for the project, or browse the AI Teammate gallery. Asana's Project Analyzer, listed in its October 2026 release notes, can recommend one from the work already in the project. Two prebuilt roles fit this job: Project Manager (its listed examples include flagging everything blocked or at risk of slipping) and Insights Analyst (its examples include preparing a briefing document for tomorrow's meeting). This walkthrough uses Project Manager.
  2. Write its guidance. Asana's setup is define the role, give it guidance, add skills, grant access. Keep the guidance short and concrete (the template is below).
  3. Grant access to one project only. A teammate doesn't get all your access by default; you share access with it. It can't see private projects unless it's added to them. Give it the renewals project and nothing else for the first month.
  4. Leave email unconnected at first. Teammates can work with Gmail, Outlook, Google Calendar, Outlook Calendar, Slack, Teams, Google Drive, OneDrive, Canva and HubSpot. But Asana's community team said a teammate can read or act on any email the connected user can access in Outlook, with no setting in Asana to limit it to one domain. For a broker holding client claims history, that is a month-two decision.
  5. Add a skill for your agenda format. Editors and admins can add skills from the Skills Library on the teammate's Skills tab, or upload a Markdown (.md) file to create a custom skill. A one-page "renewals agenda" skill keeps the output the same shape every week.
  6. Share it and set roles. A teammate is private to its creator until shared. Customising one needs Admin or Editor access; someone with User access can use it but can't change its settings. Make the office manager the only editor.

The guidance the broker used

Role: You prepare and follow up the Thursday renewals review.

Prep (when assigned "Prep: renewals review"):
- Read the Renewals project. List policies renewing in the next 60 days.
- Order by renewal date, then flag the five largest premiums (Premium field).
- For each: renewal date, handler, open tasks, anything overdue,
  anything blocked and what it's waiting for.
- Post the agenda as ONE comment on the prep task. Max 300 words.
- Do not contact clients. Do not change any task.

Follow-up (when @mentioned with meeting notes):
- Create one task per action, with assignee and due date from the notes.
- If an owner or date is missing, list it in a comment and ask me.
- After finishing, remove yourself as a collaborator from tasks you created.

Never reply to comments unless you are @mentioned.

The last two lines matter more than they look. Both come from a failure covered further down.

How do you get an Asana AI Teammate to prepare a weekly meeting?

To get an Asana AI Teammate to prepare a weekly meeting, create a recurring task, "Prep: renewals review", due every Wednesday at 2pm and assign it to the teammate. Asana's community team named this pattern directly: for an action that should repeat, assign a recurring task to an AI Teammate. When the task comes round, the teammate does the work, posts the agenda as a comment, and that run costs one request.

An illustrative agenda from week two:

Renewals review, Thursday (14 policies renewing in the next 60 days)

Largest five by premium: commercial combined for a joinery firm (renews in 19 days; awaiting claims history from client, 6 days overdue); fleet policy for a courier (renews in 33 days; two of three insurer quotes received); shop and contents for a florist (renews in 41 days; no open tasks); professional indemnity for an architect (renews in 48 days; proposal form not yet sent); landlord policy, four flats (renews in 52 days; no open tasks).

Blocked: joinery claims history (waiting on client). Overdue: 3 tasks across 2 handlers.

What the office manager corrected that week: the florist's renewal had been moved to a different insurer by phone and nobody updated the task, so it shouldn't have been listed as "no open tasks". That's a data problem, not an AI one, and the fix was a team rule that renewal decisions go into the task the same day. She also added one line to the guidance: "If a renewal is under 21 days with anything overdue, put it first", because the agenda had ordered by premium when the real priority was the joinery policy, 19 days out with documents missing.

One-off prep is a Dash question, and usually free

Not every meeting deserves a teammate. Before an unscheduled call with the courier about its fleet renewal, the director simply asks Dash: "What's open on the courier fleet renewal, and what did we decide about it in the last two renewals reviews?" Asana's AI Meetings page describes this use: you can ask Dash about previous decisions and open topics and get answers based on meeting summaries and connected work. Asana's community team listed answering questions about work in Asana and finding information among the things that don't use a request, so this kind of lookup normally costs nothing. Keep the teammate for the weekly, repeatable job and let people use Dash for the rest.

If you want a deeper brief for a single client meeting, pulling from a CRM and email rather than one project, the one-page client meeting brief covers that build.

How do meeting notes become follow-up tasks with an Asana AI Teammate?

Meeting notes reach an Asana AI Teammate in two ways, depending on your plan: AI Meetings capture on Advanced, or pasted notes with an @mention on Starter.

On Advanced: AI Meetings does the capture

AI Meetings, which Asana announced as available in November 2026, transcribes Zoom, Google Meet and Microsoft Teams calls through the Asana Desktop app, or imports Zoom cloud recordings of meetings you host. It shows a consent reminder before recording, and you're responsible for getting consent from everyone on the call. Suggested tasks appear as drafts at the bottom of the meeting; you edit, accept or decline each one, and Asana says it may suggest assigning some to an AI Teammate. One request covers a transcript up to 60 minutes, and each further hour or part of one costs another. Capture from a browser isn't supported, so whoever runs the meeting needs the desktop app, or the Zoom integration for Zoom meetings they host.

On Starter: paste the notes and @mention the teammate

Assigning a task or @mentioning a teammate in a comment both start work. After the meeting, the chair pastes the bullet notes into a comment on that week's prep task and writes "@Renewals PM create the follow-up tasks from these notes". Notes from an outside tool work too; sending Fathom or Fireflies notes into Asana with Zapier shows how to automate that hop.

A before and after, using week two's notes (illustrative):

Notes pasted in: "Joinery: commercial handler to call client re claims history, today. Courier fleet: chase third insurer quote, need by Mon. Architect PI: send proposal form this week. Landlord: fine. Florist: confirm moved insurer, close task."

Teammate's comment back: "Created 4 tasks: Call joinery client for claims history (commercial handler, due today); Chase third fleet quote (personal lines handler, due Monday); Send PI proposal form to architect (no owner in notes, assigned to you, due Friday); Close florist renewal task (owner?). Waiting for your approval to assign the fleet task: the personal lines handler isn't on this task yet."

That last line is a built-in checkpoint. Asana's help material describes approval gates for certain actions, such as assigning work to someone who isn't already involved with the task or deleting a task the teammate didn't create, and Asana's engineering team says teammates must get explicit approval before any privacy-sensitive action. Approve it, then fill in the missing owner. Two minutes, and every action from the meeting now has a name and a date. For more ways to turn notes into tasks with other tools, see turning meeting notes into tasks automatically.

How do you chase overdue follow-ups with an Asana AI Teammate without nagging?

To chase follow-ups without nagging the whole office, create a second recurring task, "Chase: renewals follow-ups", due Monday, Wednesday and Friday at 9am, assigned to the teammate, with this instruction in its description:

Find tasks in Renewals created from a meeting in the last 14 days.
For each task due in the next 2 days or already overdue:
- post one short comment to the assignee with the due date and
  the renewal date it protects.
Do not comment on the same task twice in one week.
List everything you chased in one comment on this task.

Three things to know about how this behaves:

  • Replies come back through the task. If someone replies to the Asana notification email, the reply becomes a task comment and the teammate can react to it. A standalone email sent outside that thread isn't processed.
  • It shouldn't chase clients. With Gmail or Outlook connected, a teammate can act in email. For a broker, client-facing chasers about claims history or proposal forms should go out from a person. Ask the teammate to draft the client chaser in a comment, and the handler sends it.
  • Check the nudges actually land. In week one, ask each handler whether the teammate's comments reached their Asana inbox. If someone isn't following the task, a comment can go unseen.

What early mistakes should you avoid with Asana AI Teammates?

The most expensive early mistake with Asana AI Teammates is a teammate that answers everything. A user on Asana's community forum described exactly this in October 2026: a teammate built to turn meeting transcripts into tasks became a collaborator on each task it touched, so every comment from the team woke it up, and it drained a 25-request allowance quickly. A polite instruction ("do not reply unless mentioned") didn't stop it. What worked was tighter trigger conditions and telling the teammate to remove itself as a collaborator once finished; afterwards an update that had used several requests used one. That's why the broker's guidance ends the way it does.

Other mistakes to watch for:

  • Cancelled work treated as a prompt. An Asana partner reported in early 2026 that rejecting or cancelling a teammate's task could read to the teammate as a request to try harder. If you want it to stop, say so in a comment and unassign it.
  • Too much access, too early. Giving the teammate a whole portfolio "so it has context" means its agenda pulls in projects nobody asked about. One project, one meeting, one month.
  • A paused recurring task nobody notices. If requests run out mid-month, Wednesday's prep silently doesn't happen. Put the office manager on the 65% alert email.
  • Expecting it to build automations. A teammate can't create rules for you. Dash can help set up or troubleshoot rules; a teammate can be assigned work by one.

What can an Asana AI Teammate see, and what should the team agree first?

An Asana AI Teammate's effective access is always bounded by the permissions of the person who triggers it, so a junior handler can't use it to see a director's private project. Teammates work inside Asana's permission model, and a teammate's memory follows the same rule: a memory is only visible to people who can view the task where it was formed. Every action is recorded in AI action logs that show tasks created, comments posted, searches run and objects changed, and the results are visible to anyone with access, not only the person who triggered it.

On the provider side, Asana says its AI partners, OpenAI and Anthropic, don't train on customer data and don't retain it after a query, and that those partners currently only have servers in one country. If your clients' policies or your own compliance rules restrict where personal data is processed, settle that before the teammate reads a single claims history. The weekly client status report tutorial has a section on deciding what client data belongs in AI drafts at all.

Agree three house rules in writing: the teammate never contacts clients; it never deletes or closes a task (only people do); and anything it creates gets checked by the meeting chair within a day.

Is the AI Teammate earning its requests after two weeks?

You can tell whether an Asana AI Teammate is earning its requests by looking at four things after two meeting cycles.

  1. The AI action log. Open the teammate's activity and scan every action. Anything it did that nobody asked for is a guidance problem; fix the instruction, not the people.
  2. Its memory section. Each teammate profile shows what it has retained. Delete anything wrong (for example, "the florist renewal is open") so it doesn't carry an error forward.
  3. Requests used. Admins can see usage in the Admin Console under the Asana AI tab, then AI Requests; it updates in real time. Individual users have no personal dashboard. The broker's target was 22 or fewer per month.
  4. Follow-ups closed on time. Count the meeting actions due in the fortnight and how many were done by their date. An illustrative before-and-after: 9 of 16 before the teammate, 14 of 17 after, with the chase comments doing most of the work.

If the numbers are good, the next step is adding a second meeting, not a second teammate. If they aren't, the cause is almost always the project data feeding the agenda, and the fix is a five-minute Wednesday tidy before the prep run.

If you want help fitting a teammate into how your team actually runs its meetings, I offer one-to-one AI consultations for small businesses.

Asana AI Teammates: questions small teams ask

Do AI Teammates cost extra on Asana Starter?

Not as a separate fee: they come with every paid plan, but each successful run uses an AI request. Starter and Advanced include 5 AI requests per user per month, capped at 50 per account, shared between AI Teammates and Asana Dash. Extra requests cost $0.50 each in prepaid packs or $0.60 on monthly pay as you go, and admins can set a spending cap.

Can I stop a teammate reading every email in a connected mailbox?

Not from inside Asana. Asana's community team said in October 2026 that a teammate can read or act on any email the connected user can access in Outlook, and suggested a Microsoft admin restrict the integration to specific users or groups. The simpler control for a small team is not connecting a mailbox until the meeting workflow has run cleanly for a month.

What happens when the account runs out of AI requests?

Request-based actions pause until the monthly allowance resets or extra usage is turned on. Asana's community team said a recurring task assigned to a teammate showing paused is usually a reached request limit. Admins get email alerts at 65%, 85% and 100% of the allowance, and those thresholds can't be changed.

Can a teammate join the meeting and take notes itself?

Not as a teammate. Recording is a separate feature, AI Meetings, on Advanced and above, which transcribes Zoom, Google Meet and Microsoft Teams calls through Asana Desktop and suggests tasks you accept or decline. It uses one AI request per transcript of up to 60 minutes, and suggested tasks can go to people or to an AI Teammate.

What are Asana AI Teammates?

Asana AI Teammates are prebuilt AI agents, with 30-plus ready-made roles or your own, that you assign tasks to like a colleague. They read and write comments, take context from the task, and stay within the permissions of the person who triggers them. They are available on paid plans, and each successful run uses one AI request.

How many AI requests does a weekly meeting use?

Each successful run uses one AI request. In the six-person broker example, a weekly agenda run and a weekly notes-to-tasks run use 4 to 5 requests each a month, while a chase every weekday would add about 22 more, which is why the schedule is worth planning on paper before you switch anything on.

Further reads

Sources: Asana pricing page and FAQ; Asana AI, AI Teammates, Dash and AI Meetings product pages; Asana engineering article on how AI Teammates handle context and access; Asana community forum Q&A from the Q3 2026 release webinar, October 2026 release notes, the AI Meetings launch post and user threads on teammate behaviour (all checked 17 November 2026).

Want a second pair of eyes on your setup?

Bring the process you're trying to fix. On a 1:1 call we'll work out which AI tools should handle it, which parts stay manual, and how to roll it out without disrupting the team.

Book a 1:1 call with me