SoftTech AI Registry
Asana · AI Studio

Smart Workflows

Rule-based Asana automations built in AI Studio in which a configured trigger runs a rule, an AI step evaluates conditions or generates output from selected work context and natural-language instructions, and configured Asana rule actions consume that output to persistently change work.

Recorded characteristics

Function
Asana documents Smart workflows as "rule-based automations that use AI to analyze and process work based on natural language user instructions", built in the no-code AI Studio builder and combining large language models with Asana's rules engine. A workflow has the same structure as an Asana rule: a trigger, optional conditions and actions. AI can be applied to the condition (a "conditional check by AI", where Asana AI evaluates the condition itself rather than fixed criteria), and to actions through the "Use AI" variable, which generates context-aware output such as a drafted comment, description, task or selected field value. Asana states causality explicitly in its troubleshooting article: AI Studio is built on top of Asana's rules engine, so an action must be created in the rule before anything can happen; instructing AI to comment does nothing unless an Add Comment action exists and the Use AI variable is placed in the comment field. Triggers are documented as identical to Asana's standard rule builder triggers. Workflows are published and then run on their configured trigger without a person prompting AI each time.
Data access
By default a Smart workflow can read only (a) the task that triggers the rule and (b) the project the rule lives in and the tasks in that project. Documented default task data includes title, description, assignee, due/start date, custom fields with values, type, approval or completion status, creation and last modified time, projects, sections, creator, comments, collaborators, health score and attachments on the input task (attachments require the "Include attachments added to tasks" setting). A second setting adds task activity and linked tasks (parent, dependent/blocking, subtasks). Builders extend scope by @mentioning tasks, projects, teams, portfolios, goals, users, or pasting a universal workload URL, and by attaching files to the instructions. Project @mention scope is configurable: Let AI Decide (0-150 relevant tasks), Project names only, Similar tasks (up to 10), Recent tasks (up to 200). Team @mentions expose up to 200 member names, team admins, job title and member custom fields. Asana's troubleshooting article states that when multiple projects are @mentioned the system includes as many tasks as fit the context window, up to 200, sorted by last modified date. Attachments can be read from direct uploads and from Google Drive, OneDrive, SharePoint, Dropbox and Box; the workflow accesses only the specific attached files (or relevant files inside an attached folder), attaching does not change platform access controls, and AI access to an individual file can be removed. Web access (early access, Advanced annual, Enterprise, Enterprise+, not available in HIPAA workspaces) allows real-time web search and reading of user-provided links when an admin enables it and a builder opts in per workflow.
Actions
Can take actions
External actions
Unknown
Human confirmation
Conditional
Permission basis
User permissions
Administrative control
Super admins must enable Asana AI for the domain before AI Studio can be used; division admins can disable AI features but cannot enable them without super admin verification. Admins check Asana AI and AI Studio status in the admin console under Billing > AI Studio. Since June 2025 AI Studio Basic is automatically provisioned in paid domains using Asana AI, so all domain users can see and use AI Studio and consume Basic credits. With AI Studio Pro, admins maintain an approved builder list (admin console AI Studio card, the Members table Add-ons column, or bulk CSV import) controlling which members consume credits; guests cannot be builders. Admins set a web-access policy for the domain, and the models offered in the selector depend on which AI providers the admin has enabled. Admins can view and export AI Studio credit usage. Builders control per-workflow data scope, attachment inclusion, project @mention scope, who may be added as a collaborator, model choice and human-in-the-loop checkpoints, and must publish the rule before it runs.
Default state
Disabled
Availability
Asana Help Center states AI Studio is available on Starter, Advanced, Enterprise and Enterprise+ plans, in three forms: AI Studio Basic (included in all paid domains using Asana AI, with monthly credit rate limits of 50,000 Starter / 75,000 Advanced / 200,000 Enterprise / 200,000 Enterprise+ per billing account), AI Studio Plus (purchasable in-product, 100,000 credits resetting monthly, additional credits in 100,000 increments or 1.2 million annual credits) and AI Studio Pro (annual, 20 million credits for the term, with a builder list controlling who consumes credits). AI Studio is documented as usable by customers on Asana's HIPAA offering, although web access is not available in HIPAA-compliant workspaces. Documentation notes that customers who purchased AI Studio before 16 September 2026 have different features and credit structures. This is current general availability documentation, not beta.
Licensing
Credits, not seats, meter execution. Each workflow run consumes credits based on the model's multiplier, the volume of input data, the volume of output and how often the rule runs; workflows using web access consume additional credits per execution. Asana recommends lighter models, trimming attachments, capping output length and adding conditions to control credit spend. Credits are a billing mechanism and are documented separately from permission to build or run workflows.
External model or provider
Asana states AI Studio connects to models from its AI partners (including OpenAI, Anthropic, Google, and open-weight models served through Fireworks.ai) and that Asana does not create or train these models. A model is chosen per rule. The documented default is Claude Sonnet 4.5; if Anthropic-powered features are disabled by an admin, GPT-5 becomes the default. The AI Studio models article lists Claude Sonnet 4.5, GPT-5.2, GPT-5, GPT-5 mini, Claude Opus 4.6, Claude Haiku 4.5, Kimi K3, GLM 5.2 and Deepseek V4 Flash. Asana states it does not use customer data for training and has contractual agreements prohibiting its AI partners from using customer data to train or improve their models, and that partners delete input and output data after each response.
Limitations and uncertainty
AI output is not itself a state change: an action must exist in the rule and carry the Use AI variable. Documented Use AI action points are adding comments, moving to a section, moving or adding to a project, setting task description, setting custom fields, creating a task, creating a subtask, creating approvals, setting assignees, setting due dates and adding collaborators; Asana calls this a non-exhaustive list, so the full action surface is not established. Use AI with assignee is restricted to objective instructions - Asana states it is not designed to assign work via AI assessment of a user's skills or traits and blocks publication until instructions are revised or Use AI is removed from the assignee field. Asana's AI Studio product page claims workflows can trigger actions and sync data across connected tools, but the current Help Center AI Studio articles reviewed do not document specific third-party write actions or their runtime credentials, so external action capability is recorded as unknown rather than yes or conditional. Runtime identity is not separately documented: Asana states a Smart workflow can only access projects and tasks the builder has permission to see, and that a builder can only grant access to work they can access themselves, but Asana does not document a distinct agent identity, and whether the triggering user's permissions apply at execution is not established. Human review is a best practice implemented by adding approval tasks or similar checkpoints, not a mandatory stage of every workflow. Workflows can fail to act when AI lacks sufficient context or instructions are ambiguous; Asana surfaces a warning that the workflow could not run an action, and users can open the task history and click Show reasoning to see an AI-generated explanation of how instructions were interpreted - this is Asana's reasoning display, not a full chain of thought or an immutable audit log. Attachment limits (images 5MB, attachments 25MB, images Claude-only, PDF images ignored, no access to parent or subtask attachments) constrain analysis. Retention, rollback, rate limiting beyond credits, and logging semantics are not established. AI Teammates, Asana Dash, Smart Chat, Smart Status and Asana's MCP/AI connectors are separately documented Asana capabilities and none of their authority is attributed here.

Evidence