New to AI Studio? A beginner-friendly walkthrough to setting up your first AI agent

With Asana’s AI Studio, it’s easy and quick to create a custom AI Agent directly in your project to automate your workflow.

Don’t believe me? Follow along here to get your own AI agent up and running in minutes.

Enable AI Studio (if not done already)

AI Studio is available for Asana Enterprise, Enterprise+, or annual Advanced plans. See AI Studio add-on and pricing for more info.

To enable AI Studio:

(Asana’s AI partners do not train or improve their AI models using your data.)

Your First AI Agent

We’ll create an AI Agent that will audit your tasks to encourage task best practices, either automatically or on demand. You can include any preferred etiquette and best practices your team or organization likes to use.

The more organized and consistent your tasks are in Asana, the more effective both you and your AI agents will be. Well-named tasks with good descriptions and proper use of other task metadata won’t just be easier for you to understand or locate, but will also provide Asana AI the context needed to understand and act intelligently on your behalf.

The AI Agent in Action

Here’s a not-so-great example task that I added to the project:

Moments after adding the task, the Task Audit Suggestions AI Studio rule ran automatically and added this comment:

That’s an example of what this AI Agent does: it audits tasks.

Creating the AI Agent Rule in a Couple of Minutes

Pick a project (or create a new one), then click Customize > Rules > Add > Create with AI Studio. (If you need more basic help with creating a rule, see Rules in the Asana Help Center.)

The rule you’ll create is straightforward and contains:

  • When trigger to run the rule either a) whenever a task is added to the project, or b) on demand for any single task or multiple selection,
  • Do this action to add a comment; the Use AI variable gives the agent the ability to place its results there, and
  • Guidance for AI prompt instructions to direct the agent how to audit the task and display its results

Go ahead and add the rule as described above and shown below. The rule should look like this when you’re done:

When you click on the Add comment action, it should look like this:

The full Guidance for AI instructions, part of which is shown in the above image, is available after you Expand below; copy in its entirety and paste in the rule builder to match the screenshot above. (Be aware that these instructions are just a quick sample for demonstration purposes. You should experiment and tweak them to a) perhaps not be so aggressive with the suggestions, and 2) include any of your team’s/organization’s preferences.)

Expand to see Guidance for AI instructions to copy/paste

You are an expert in how tasks should be conveyed in Asana and your job is to audit this task as specified below.

If a task fails to meet at least one of the best practices below, add a comment as shown below in the form of a compact bullet list with no heading. Include one bullet for each improvement you can suggest, in sorted order of most important recommendations first, and limiting to around five or so, replacing the <> bracketed elements as indicated:

  • <Short, scannable phrase summarizing issue>: <Cogent (sentence or two) detailed suggestion for improvement>

It’s ok to add comments even if you already added one previously to allow for a re-audit.

Use this checklist below of task best practices. Generate your suggestions both from this and from any other generally agreed upon Asana best practices, e.g., from the Asana Help Center.

Task Best Practices Audit Checklist

Task Title

  • Title should be actionable and start with a verb (e.g., “Create campaign budget” instead of “Campaign budget”)
  • Title should be specific and clear about what needs to be done; avoid vague titles that don’t communicate the action required

Task Description

  • Include all information needed for the assignee to complete the task
  • Use @mentions to link relevant people and related work
  • Include attachments or external links to centralize information
  • Use rich text formatting to improve clarity

Task Ownership/Collaborators

  • Every actionable task should have an assignee
  • Add collaborators for stakeholders who need visibility
  • Generally you should not be the only collaborator

Task Timing

  • Include realistic due dates for all tasks; include start dates if appropriate

Task Organization

  • Use multi-homing for tasks that belong in multiple projects
  • Break large tasks into smaller, manageable subtasks
  • Fill out and update custom fields as appropriate
  • Use task dependencies to show relationships between tasks

Task Size

  • Break work into bite-sized pieces that can be completed in minutes or hours
  • Avoid tasks that are too large

–End of Guidance for AI instructions–

Running the AI Agent

In addition to the agent running automatically whenever a new task is added, you can manually (on demand) run it by right-clicking a task and choosing Run a rule > Task Audit Suggestions, or via multi-select:

Next Steps

This is just one example of what an AI Agent in AI Studio can do; the possibilities are limitless.

  • Keep an eye out for repetitive work that may be candidates for other AI Agents.
  • Experiment as you build agents to learn just how powerful they can be.
  • Use clear organization and be thoughtful with the wording of your Guidance for AI prompt for best results.
  • Keep a “human in the loop” in your AI Agents. This example agent makes recommendations only. It leaves you in the driver’s seat to decide whether to apply any, perhaps by copying/pasting a new task title suggested in the comment.
  • Share your AI Agents across your team/organization to scale. This quick example just had you define this AI Agent in a single project, but for deployment, you may prefer to place it in a Bundle (Enterprise feature) or in a Project Template. Another approach (if you don’t have access to Bundles) is to use a project with just this one rule in it and nothing else. When you multi-home tasks from other projects there, the rule will run automatically. Just include one additional Action in the rule definition: Remove task from the project so tasks don’t accumulate there after doing the agent completes its work inside the task.
  • Watch for updates to AI Studio to stay on top of new functionality.

If you’ve followed along here, then you are well on your way to making Asana’s remarkable AI Studio a regular part of your workflow.

Thanks for reading,

Larry Berger, Forum Leader, Asana Services Partner, Trilogi Solutions

2 Likes

Related: @Emily_Roman recently posted this fully-automatic task re-namer:

I chose to keep the “human in the loop” with my AI automation here, but I wanted to link to that related approach in case it’s of interest.

1 Like