Agentic work management is Asana's term for running workflows where people and AI agents share the same plan, the same project and the same permissions. The features are real and available now, but the included AI allowance is far smaller than most teams assume, and that is the part worth understanding before you plan a rollout.
Agentic work management is a model where AI agents are members of a project rather than a separate tool you visit. The agent reads the project's context, does assigned work, and leaves its output where the rest of the team can see it.
Asana announced this positioning on 4 June 2026 at its Work Innovation Summit in London, describing the platform as an operating system for human and agent teams. The company's framing statistic: roughly 75% of knowledge workers use AI, while only 5% of organisations report meaningful productivity gains from it. The argument is that the gap is structural. AI that sits beside your work cannot see your work.
Three things are available now.
AI Teammates are agents that live inside a project. Asana ships a library of prebuilt ones covering marketing, operations and IT, plus industry variants, and a no-code builder for custom agents. They connect to Gmail, Outlook, Slack, HubSpot, Figma and Canva.
AI Studio is the no-code workflow builder. You set a trigger, write plain-language instructions for what should happen, and the workflow runs.
StackAI, which Asana acquired in May 2026, extends orchestration beyond Asana into CRMs, ERPs and custom infrastructure.
Announced as phased, arriving over the following months: Asana Dash, an assistant that produces a daily briefing and pulls actions out of meetings, Slack and email; Asana Service Management for IT, HR and facilities ticketing; Command for planning and product development; and Asana Client Management for agencies. Client Management now appears on Asana's pricing as available on Starter and Advanced, so check current status rather than relying on the launch material.
One caution. Asana's own pages disagree on how many prebuilt AI Teammates exist, listing 21 in one place and more than 30 in another. Treat any specific count as unstable.
The difference is where the output lands. A chatbot answers you privately, and whatever it produced has to be copied somewhere before it counts as work.
An AI Teammate is assigned to a project, so its output appears on the board where the team already works. It also inherits your existing Asana permissions rather than introducing a separate access model, which means the agent can only see what the person who configured it could see. In Asana's beta, 93% of AI Teammates were granted full edit access, which tells you how teams expect to use them once the permission question is settled.
AI Studio handles the coordination work that sits between a request arriving and someone doing it. Its actions fall into three groups.
For a bilingual Israeli team, that last item is not a minor feature. Intake in Hebrew, routing and reporting in English, without a person retyping anything, is a common and expensive gap.
This is where rollout plans go wrong. Asana meters AI two separate ways, and the second meter is the one that constrains you.
AI Studio runs on credits. The monthly allowance is roughly 50,000 on Starter, 75,000 on Advanced and 200,000 on Enterprise, pooled across the whole billing account.
AI Teammates and Dash run on requests, and the allowance is much tighter: 5 requests per user per month, capped at 50 requests per account on Starter and Advanced, and 250 on Enterprise.
Read that cap carefully, because it is per account, not per user. A 25-seat Advanced team does not get 125 requests a month. It gets 50, for the entire company. Beyond the allowance, Asana charges $0.50 per prepaid AI request, or $5.99 per user per month plus $0.60 per request on demand.
For context, seats are $10.99 per user per month on Starter and $24.99 on Advanced, billed annually. A team running agents on real volume will find the AI line comparable to the seat line, and it will arrive in month two rather than month one.
This mirrors what happened when HubSpot moved its agents to per-run billing. The pattern is the same across both vendors: the capability is priced separately from the seat, and the default allowance is a trial, not a budget.
Old way. A request arrives by email. Someone reads it, decides who owns it, opens a project, copies the detail across, chases the missing fields, and writes a status update on Thursday so the weekly meeting has something to look at. The coordination cost is invisible because it is spread across everyone.
New way. The request hits a form that AI Studio validates, classifies and routes. An AI Teammate drafts the brief from the attached material and flags what is missing. The status update writes itself from what actually moved. People spend their time on the judgment calls rather than the handoffs.
The gap between these two is not a feature gap. It is an architecture gap. Agents only produce useful work when the underlying project structure, custom fields and intake are already clean. Context before automation. An agent pointed at a messy workspace produces confident nonsense faster than a person could.
Asana Client Management productizes a lot of what agencies build by hand: a two-way client portal, capacity planning, margin tracking, and four agency agents covering account health, creative production, profitability and resourcing.
If you have assembled that yourself out of custom projects and rules, the calculation changes. The question is no longer whether to build it. It is whether your configuration is better than the product, and whether the plan you are on supports the volume you would actually run.
Start with the readiness question, not the agent catalogue.
Asana's agentic features are genuinely useful and genuinely constrained. The capability is available today on paid tiers; the included allowance is small enough that any serious deployment needs a cost model before it needs an agent. Teams that audit their workspace and model consumption first get a working system. Teams that start by switching agents on get a bill and a mess.
IV-Lead is Israel's first Asana Solutions Partner and a HubSpot Solutions Partner (Gold), which means we plan these rollouts with both meters in view, including what happens where Asana and HubSpot both charge for AI against the same process.
Book a 30-minute Asana plan-fit call and we will size the workflow, the tier and the AI consumption before you commit to either.