The Asana AI Playbook: What's Verified, What's Not

What Asana AI is and which plans include it

Asana's pricing page lists Asana AI (AI Teammates, Asana Dash and AI Studio) as included from Starter ($10.99 per seat a month billed annually), Advanced ($24.99 per seat a month billed annually) and Enterprise^[1]; the free Personal plan (2 users) doesn't include it. You can test it on Asana's free Advanced trial (no credit card required). What varies by tier is the usage allowance: 50,000 AI Studio credits a month on Starter, 75,000 on Advanced, and AI Teammates and Dash requests capped at 50 per account (see the pricing mechanics below).

As of its 4 June 2026 Work Innovation Summit in London, Asana repositioned itself around "agentic work management" — explicitly framing the platform as an operating system for human and agent teams, not a project tracker with a help bot bolted on.

Asana's own justification for the push is a stat worth sitting with: it claims roughly 75% of knowledge workers already use AI, but only 5% of organizations report meaningful productivity gains from it. That's Asana's own gap analysis, and it's the stated reason for the pivot.

The build-out behind that pivot is concrete and named:

• AI Teammates connect natively to Gmail, Outlook, Slack, HubSpot, Figma, and Canva, with a no-code builder for custom agents.
• Asana acquired StackAI in May 2026 specifically to extend agent orchestration beyond Asana itself — into CRMs, ERPs, and custom infrastructure.
• Asana's own marketing can't agree on how many prebuilt AI Teammates exist: one page lists 21, another lists "more than 30." Treat the discrepancy as a sign the product is shipping faster than the marketing copy describing it, not as a red flag on the agents themselves.

Key takeaway — AI Teammates and AI Studio are named, documented features with real integrations, not just a landing-page promise^[2]. The support-chatbot caveats apply to the chatbot, not to the rest of Asana AI.

What this playbook can't confirm is quality. The user-review evidence behind it is five reviews, and the one AI touchpoint they describe in detail is a customer-support chatbot that handles billing and account questions. Beyond the AI features its pricing page lists from Starter, the quality of what Asana claims about in-app AI for project and task management is unverified in this evidence set — not false, just not something we can confirm or deny here. Treat it as something to pilot and test on your own account, not a feature to plan your workflow around sight unseen.

What AI Studio actually automates

AI Studio's workflows split into three documented categories, and they map to genuinely different risk levels:

Category	What it does	Examples
Intake	Filters and routes incoming work	Duplicate detection, SLA-based routing, classification/scoring
Surface	Flags things humans would otherwise miss	Risk flags, stakeholder roll-ups, document/web search
Act	Takes action or generates output	Drafting briefs/comms from templates, cross-tool syncing, language translation

Intake and Surface are low-stakes — they're filtering and flagging, not changing anything. Act is where the 93%-full-edit-access finding (see governance below) becomes relevant: an agent that generates a client-facing brief and syncs it across tools is also an agent that can push a wrong draft live if it's not scoped down first.

Pricing mechanics: credits, requests, and the hidden overage math

Asana meters AI two separate ways, and the two meters don't resemble each other — this is the detail most teams miss when budgeting.

Tier	AI Studio credits/month	AI Teammates/Dash request cap (per account)	Seat price
Starter	50,000	50 total	$10.99 per seat a month billed annually
Advanced	75,000	50 total	$24.99 per seat a month billed annually
Enterprise	200,000	250 total	Contact sales

The request cap is the trap. It's quoted as "5 requests per user per month," but it's hard-capped account-wide at just 50 total requests on Starter/Advanced^[1] — meaning a 20-person team shares the same 50-request pool a 10-person team gets (5 requests a user hits the 50 cap at 10 users). Burn through it and overage runs $0.50 per prepaid request, or $5.99/user/month plus $0.60/request on an on-demand plan.

Heavier AI use pushes costs further. One reported case: a team needing "meaningful AI automation" moved to AI Studio Plus at $150/month for 100,000 credits, then burned roughly 200K credits per person monthly anyway — adding ~$300/user/month on top of the Advanced license ($30.49 per seat a month billed monthly), for an effective ~$330/user/month on AI-heavy workflows. A modeled total-cost-of-ownership comparison for a 50-person team put Asana at $17,994/year (Starter subscription for 50 seats $6,594 a year + labor $9,000 + error/rework $2,400) against $17,400/year for Notion in the same model — close enough that AI usage intensity, not list price, is likely the deciding variable.

Two structural pricing issues compound this:

• Seat increments force overbuying. Past 5 users, Asana sells seats in blocks of 5 (jumping to 25-seat blocks at 100+ users on Advanced). One forum-reported case paid for 24 empty seats at roughly $21/month each — about $500/month wasted on seats nobody used.
• SSO is Enterprise-only. SAML SSO isn't available on Starter or Advanced at all; it requires upgrading to Enterprise pricing, reported around $30–40+/user/month. If your security team requires SSO, factor that tier jump into the AI cost conversation now, not at renewal.

Key takeaway — Budget AI Studio credits and AI Teammate requests as two separate line items with two separate overage risks. The request cap in particular is account-wide, not per-seat — a handful of power users can exhaust it for the whole team by mid-month.

Governance: permissions, model provenance and data handling

One governance detail matters more than the feature list: in Asana's own beta testing, 93% of AI Teammates were granted full edit access rather than restricted permissions. If you're piloting AI Teammates, don't accept the default. Explicitly scope each agent to read-only or limited-edit access before it touches live project data^[3] — the beta data suggests most teams won't do this unless they're told to.

Whose AI is this, actually?

Asana doesn't disclose, in any of the material available here, which model or vendor powers its AI features. We don't know if the support chatbot or the in-product Asana AI runs on a proprietary model Asana built in-house or routes to a third-party LLM provider.

This matters for two practical reasons. First, if you're in a regulated industry (healthcare, finance, legal), your compliance team will ask which model, which data residency, which retention policy before approving AI use at all — and none of that is answerable from public evidence alone. Second, unclear provenance usually means unclear data handling.

Ask your Asana account rep directly, in writing, before you feed anything sensitive into any AI-labeled feature. Get the answer on your contract or in an email, not a sales call.

Data handling: what to get in writing

On training, Asana's AI page is explicit: Asana's AI partners don't use your data to train their models and are required to delete it after each query^[2].

Two data questions remain open in this evidence set:

• Who is data shared with (subprocessors, model vendors)? Not disclosed; there's no subprocessor list here.
• Is there an opt-out? Not disclosed.

None of this means Asana is doing something wrong. It means the documentation needed to confirm good practice wasn't available for this review. If you manage a workspace with client names, financial figures, or anything under an NDA, don't assume AI features are isolated from your data until you've seen it in writing.

Key takeaway — "Not disclosed" is not the same as "safe." Before you route sensitive task content through any AI feature, get Asana's data-handling answer in writing — don't infer it from the marketing page.

The support chatbot: where it fails and when to escalate

Asana runs an automated chat layer for customer-service inquiries — billing questions, account deletion requests, general account troubleshooting. It's the one AI touchpoint multiple independent reviewers describe in detail.

And the reviews are not kind. The consistent pattern across them: the bot loops without resolving the actual issue, especially on two high-stakes requests — refunds and account/data deletion.

This is useful information precisely because it's specific. If your query is a simple account question (password reset, where's my invoice, how do I update a seat count), the bot may route you fine. If your query involves money moving or data being permanently removed, budget zero trust in it.

How to use the chatbot without wasting time

1. Open with the exact issue type up front. Lead with "billing dispute" or "account deletion request" in your first message rather than general small talk — this won't fix the bot's limits, but it shortens the loop before you escalate.
2. Set a two-reply limit for yourself. If the bot hasn't routed you to a human or resolved the issue within two exchanges, stop. Reviewers describe looping as the dominant failure mode — more messages rarely break the loop.
3. Screenshot everything. If you later need to dispute a charge with your bank or card provider, a timestamped record that you attempted resolution through official channels strengthens your case.
4. Never treat a chatbot confirmation as a deletion or refund confirmation. Verify independently (email confirmation, account status check) before assuming the request processed.

Escalate past the bot immediately for anything high-stakes

This is the single most actionable piece of advice the review evidence supports, and it's worth treating as a rule rather than a suggestion.

For billing disputes: Don't cycle through the chatbot hoping for a different outcome. Go directly to your payment provider (card issuer, PayPal, etc.) and file a formal dispute if Asana's support channel isn't resolving it. This is the explicit workaround reviewers point to, and it sidesteps a bot that isn't built to handle contested charges.

For account/data deletion requests: Same logic. If the chatbot can't confirm deletion, pursue it through a documented support ticket or email to Asana support that creates a paper trail, rather than relying on chat confirmation alone.

For anything involving legal or compliance deadlines: Don't let a chatbot interaction be your only documented attempt at resolution. Escalate early enough that you have time for a human-reviewed outcome before a deadline hits.

The most valuable AI skill in Asana support today might be knowing exactly when to stop using it.

Role-by-role: how to apply this

If you're the admin managing the company's Asana subscription

Your job is risk reduction, not feature maximization. Before renewal, confirm in writing (not from the chatbot) what your actual billing terms are, your seat count, and your cancellation policy. Don't let a support-chat exchange be your only record of a pricing conversation — get it in an email thread you can reference later.

If you're evaluating Asana AI for your team

Treat it like any vendor claim you haven't tested: pilot before you commit. Pick one project, turn on the AI features your tier includes, and run it for two to four weeks against the same workflow your team already does manually. Compare actual task output, not Asana's marketing description, and only roll it out further once your own team confirms it holds up.

If you run an agency or a marketing team

One third-party assessment characterizes Asana AI as suited to marketing teams that need formal, recorded approval workflows — a documentation and sign-off tool — rather than a tool that generates creative deliverables directly. That lines up with where Asana has actually expanded next: Asana Client Management, now available on both Starter and Advanced, adds a two-way client portal, capacity planning, margin tracking, and four dedicated agency agents (account health, creative production, profitability, resourcing) — built for agencies managing approvals and billable work, not for teams wanting AI to produce the creative work itself.

If you're a customer trying to get a refund or delete your account

Skip the chatbot for the core ask. Use it only to open a ticket or confirm you're a registered account holder, then push for human support or go straight to your payment provider if money is involved. Document every step with timestamps.

A quick reference table

Touchpoint	Status	What the evidence shows	Recommended move
Asana AI (AI Teammates, Asana Dash, AI Studio)	Listed on Asana's pricing page from Starter; quality unverified in this evidence set	No review data confirms quality or reliability	Pilot on one non-critical project before depending on it
AI Teammate permissions	93% granted full edit access in Asana's beta	Default access is broad	Scope each agent to read-only or limited edit before it touches live data
Support chatbot (billing, account questions)	Verified by multiple reviews	Loops, frequently fails to resolve refunds and deletion requests	Use for simple queries only; escalate to human support or payment provider for anything high-stakes
Underlying AI model/provider	Not disclosed	N/A	Get written confirmation from your Asana rep before sharing sensitive data
AI pricing (included vs. add-on)	Included from Starter, not on the free Personal plan (Asana pricing page)	Not covered by reviews	Check the per-account AI Studio credit and AI Teammates/Dash request limits for your tier before relying on it

ROI: what we can and can't claim

We're not going to hand you a made-up percentage here. There's no outcomes data in this evidence base — no case study, no benchmark, no customer-reported efficiency gain tied to Asana's AI. Any specific ROI number for Asana AI would be invented, and we don't do that.

What we can say honestly: the measurable win available to you right now is avoided waste, not gained output. Reviewers describe real time lost looping with a support bot that couldn't resolve their issue. The fix — a two-reply cutoff and immediate escalation — is a process change you control, and its payoff is in hours not spent arguing with a chat window, not a productivity multiplier from the AI itself.

When you pilot Asana AI features in your own workspace, measure your own before/after numbers on a real project. That internal benchmark is worth more than any industry-wide stat you'll find quoted elsewhere, because it reflects your actual workflow and your actual data sensitivity constraints.

Cost pitfalls to price in before you commit

Because Asana AI is bundled from Starter, the budgeting risk isn't access — it's the usage allowance, which varies by tier, and the overage you pay once it runs out.

Before you factor AI capability into a renewal decision or a plan upgrade:

• Get written confirmation of exactly which AI features and allowances apply at your current tier.
• Don't let a sales conversation substitute for contract language — if AI access matters to your renewal decision, get it specified in the agreement.

Key takeaway — Asana AI comes with every paid tier from Starter, so don't upgrade just to unlock it; an upgrade raises the AI Studio credit allowance (75,000 on Advanced vs 50,000 on Starter), not access.

Power moves, condensed

• Set a hard two-reply rule with the support chatbot. Escalate immediately past that point for anything involving money or data deletion.
• File payment disputes with your card provider or payment platform directly when a billing issue isn't resolved through Asana's own channels — don't wait on repeated chatbot attempts.
• Pilot, don't deploy. Test any Asana AI feature on one project with a defined measurement window before trusting it with core workflows.
• Scope every AI Teammate down from full edit before it touches live project data, starting with agents in AI Studio's Act category.
• Get data-handling answers in writing before routing sensitive content through AI features — subprocessors and opt-out aren't documented in this evidence set.

The bottom line for power users

Asana AI (AI Teammates, Asana Dash and AI Studio) comes with every paid plan from Starter; what changes by tier is the allowance, and the account-wide request cap is the line to watch. The Asana AI touchpoint user reviews document is a support chatbot that struggles with the exact situations where users need it most: money and data. That's a real limitation worth knowing before you lean on it.

The in-product Asana AI may well be more capable — plenty of software companies have thin support-bot experiences sitting next to genuinely useful in-product AI. But we can't confirm that from the evidence here, and neither should you until you've tested it yourself inside your own workspace, with your own data, against your own workflow.