Asana AI features and admin controls

Available on Asana StarterAdvancedEnterprise, and Enterprise+ tiers. Certain AI features are available on legacy tiers Premium, Business, and Legacy Enterprise. Visit our pricing page for more information.

Asana AI features use artificial intelligence (AI) to sort, filter, categorize, or otherwise analyze data and/or content to help users in your organization optimize their work.

Optimize with Asana AI

Asana AI refers to two sets of features:

  • Powered by Asana ML: this feature set is powered by Asana’s own proprietary machine learning models.
  • Powered by AI partners: this feature set relies on generative AI and large language models (LLMs) developed by third parties (e.g., OpenAI, Anthropic).

Customers can choose to enable Asana AI features for their organization or workspace. If Asana AI features are not enabled:

  • The users in the organization/workspace will not have access to the AI features.
  • The organization/workspace's data will not be processed to power these features.

Read further to learn more about Asana AI features and the safeguards we’ve put in place for our customers.

Please note: Customers using Asana's HIPAA offering can enable Asana AI features powered by Anthropic. Asana AI features powered by OpenAI are not available for HIPAA customers.

Data used for Asana AI features powered by Asana ML

Asana AI features powered by Asana ML use metadata from your organization/workspace to help users optimize their work with AI.

Metadata does not include user generated content (e.g., task titles and task descriptions).

Data used for Asana AI features powered by AI partners

Asana AI features powered by AI partners use metadata and user generated content from your organization/workspace (e.g., content you enter into Asana like task titles and task descriptions, which may include user names) to help users optimize their work with AI.

AI partners do not use your data to train or improve their AI models. AI partners are required to configure their services to promptly delete your data following completion of any query and/or response.

To learn more about how Asana processes your information, please see Asana’s Privacy Statement.

AI-powered features administered by the Asana AI admin control(s)

The features below can be enabled/disabled by the Asana AI toggle in the admin console.

  • Recommended projects: In the Home tab, under the Projects widget, a list of Recommended projects can be displayed. These recommendations are generated using a collaborative filtering machine learning model, which infers relationships between users and projects based on which Asana projects a user and their frequent collaborators interact with. If Asana AI features are disabled in the Admin Console, users will only receive a generic list of recommended projects in Asana.

  • Priority inbox: In Inbox, a new Sort option for inbox activity is displayed. This allows users to sort their inbox by Relevance.

  • Task recommendations: Shows users personalized recommendations of tasks to prioritize.

  • Recommended teams: In the Team Directory, using the Recommended sort, a list of recommended teams is displayed. These recommendations are generated using a collaborative filtering machine learning model, which infers relationships between users and teams based on which Asana projects a user and their frequent collaborators interact with.

  • Top projects and portfolios: In the Sidebar, using the Top sort in the Projects section, projects and portfolios are displayed in a ranked list. This list is generated based on the projects and portfolios that a user has interacted with. If Asana AI features are disabled in the Admin Console, users will only receive a generic list of top projects in Asana.

The following features can be enabled/disabled by the Powered by AI partners control in the admin console.

  • Smart editor: Write clearer, more compelling responses that strike the right tone.
  • Smart summaries: Available for tasks, projects and portfolios. Use smart summaries on tasks to get action items and highlights from conversations, tasks, and comments without another meeting. Use smart summaries on projects and portfolios to keep up to date with an easily scannable summary of all recent activity and highlights in one place.
  • Smart fields: Organize projects to support cross-functional collaboration with auto-generated custom fields.
  • Smart status: Identify blind spots, open questions, and roadblocks to find an efficient path towards your goals with automatic status updates.
  • Smart answers: Get timely answers and insights about projects, identify blockers, and determine next steps.
  • Smart goals: Standardize and improve the effectiveness of goal-setting with smart goals. Smart goals analyzes existing goals and offers suggestions for improvement, including adding success criteria, impact summaries, parent goals, supporting subgoals, and assigning accountable teams.
  • Smart projects: Quickly create customized projects with smart projects. Smart projects suggests descriptions, sections, and custom fields all based on the project's name and purpose.
  • Smart rules: Create rules with AI by providing simple, and natural language instructions via prompts.

Enabling and disabling Asana AI features 

Super admins of organizations and workspace billing owners and creators can enable or disable Asana AI features as follows:

Optimize with Asana AI settings.png

  • Click your profile picture in Asana and navigate to the Admin console
  • Click the Settings tab
  • Under Domain settings select Optimize with Asana AI

From this window you can choose to enable or disable Asana Machine Learning for your entire organization or workspace by checking or unchecking the appropriate box.

By disabling Asana Machine Learning, your organization’s data will not be used to inform artificial intelligence features and these experiences will be disabled for the entire organization.

If you are an admin of an organization that does not have any super admins, you can disable these features, but cannot enable them again. To enable them, you will need to complete the super admin verification process or contact the Support team. A warning banner will be displayed in-product before an admin makes this change. This also applies to divisions in Asana as divisions do not have super admins. In order to enable Asana AI features for a division, a division admin would need to complete the super admin verification process and become a super admin of the organization. 

Embeddings

Some Asana AI features require embeddings to work, while others use embeddings to improve performance. Leveraging embeddings generally works in two key phases:

  1. Creating embeddings
  2. Generating responses

At this time, domains that have enabled Asana’s HIPAA compliance option or utilize EKM are not eligible for embeddings.

What are embeddings?

Embeddings are numerical representations of work in Asana that capture the similarities between objects, regardless of how those objects are named or organized. Embeddings help Asana's algorithms better understand relationships between tasks, projects, portfolios, goals, and all other work in Asana. This produces far more accurate and high-quality results for experiences involving search, retrieval, and recommendations.

How are embeddings created?

  1. Data from your domain is sent to an embeddings model created by AWS OpenSearch, hosted by Asana. In the case of Smart Fields only, data is instead sent to an embeddings model created and hosted by OpenAI.
    • OpenAI is an AI subprocessor who does not store or train on your data (data is only used to generate the embeddings).
  2. Asana stores the embeddings mappings in AWS OpenSearch, which Asana queries to power search-like and information retrieval experiences. Note: AWS OpenSearch is the same infrastructure that powers Asana’s standard search today.

How are embeddings used in practice to generate responses?

Asana will use embeddings in a variety of use cases that will help users work more efficiently, such as improved search speed/accuracy, more tailored recommendations, logical grouping of alike objects. A specific search related use case might be finding and retrieving information using natural language prompts.

For example, a user may ask Asana AI to “Find all of the tasks and subtasks I am on with John Doe where we’ve discussed the branding guidelines for our Sept 2024 marketing launch”. Semantic search queries like the one expressed above are only made possible through the use of embeddings.

An example flow of data with embeddings enabled might look like following:

  1. Asana receives a question from a user.
  2. Asana rephrases the question for optimal responses from various systems.
  3. The rephrased question is passed to an embeddings model, which finds a list of relevant objects based on the question, and sends a response back to Asana.
  4. The responses are analyzed by Asana.
  5. Asana displays the relevant response to the user.

Each step may invoke a call to our subprocessors.

How is my data protected?

You can learn more about Asana’s Security and Privacy practices at our Trust Center.

You can learn more about OpenSearch’s security practices here and about AWS here.

Proactive AI

What is proactive AI?

Proactive AI is a capability that will surface AI features to relevant users at the optimal time, without them needing to manually request it. An example of proactive AI is sending a project smart summary insight to project members when certain types of relevant activity have occurred, without them having to explicitly ask for it.

Proactive AI will use data from your organization - including metadata and user generated content - to identify the right moment to surface AI features and insights, and the right users to surface them to.

How does proactive AI work?

There are a variety of Asana AI features that are proactive, but the general flow is:

  1. Users create and edit work content (such as tasks or projects) throughout their regular use of Asana. These activities are aggregated and analyzed to identify high-level patterns about how work is progressing.
  2. When certain patterns are recognized or conditions are met, Asana AI will determine an optimal time to trigger a proactive AI feature and identify the most relevant users to surface the feature to.
  3. Users are made aware of the proactive AI feature through a notification or UI element.
  4. They can choose to take a recommended follow up action, sign up for similar recurring outputs on an ongoing basis, and/or dismiss the proactive AI.

For example, a user may receive a proactive project smart summary insight inside Asana because the project meets certain conditions like reaching a certain amount of recent activities. The user(s) who receive the proactive insight are selected based on a variety of relevancy factors, such as users recently added to the project or project admins, to ensure that proactive features are relevant to the right people at the right time.

After experiencing the proactive AI feature, the user can optionally choose to take a follow up action, such as opting into this insight on a regular recurring basis or copying the insight to share it with others.

Will outputs triggered by proactive AI count against my AI usage limit?

No - outputs generated by proactive AI features or insights will not count against your organization’s AI usage. However, some proactive features will recommend follow up actions or allow users to opt into that feature on a continuous basis. If a user takes a follow up action on a proactive AI feature, then the initial proactive output will count towards AI usage, and any follow up outputs that are generated from the recurring opt-in will count towards your AI usage as well.

How is my data protected with proactive AI?

Proactive AI leverages the same security and privacy practices as the rest of Asana’s AI features. There are no changes to the Asana AI safeguards we have in place with respect to your data.

You can learn more about Asana’s security and privacy practices at our Trust Center.

Feature breakdown by provider

Feature Provider
Smart answers - ask questions Anthropic
Smart answers - suggested questions OpenAI
Smart status Anthropic
Smart fields OpenAI
Smart editor OpenAI
Smart summaries on tasks OpenAI
Smart summaries on projects Anthropic
Smart summaries on portfolios Anthropic
Smart goals Anthropic
Smart projects OpenAI
Smart rules

OpenAI

AI chat

Anthropic and OpenAI

Asana AI limits

Visit our dedicated article for a detailed overview of Asana AI limits.

Asana’s AI safeguards

We commit to being open and clear about our partnerships, security, data collection, data protection, and management practices. Our AI partners adhere to the following guidelines: 

  • They do not train or improve their AI models using your data.
  • They must configure their services to promptly delete your data following completion of any query and/or response.

Asana customers can learn more in the Asana Trust Center, which includes the AI Fact Sheet.

Was this article helpful?

Thanks for your feedback