Asana AI: how to create specialized prompts

Have you ever asked a new colleague for help, given vague instructions and received something completely different from what you expected? Asana’s artificial intelligence works the same way. The difference between a generic result and a ready-to-use deliverable lies in the quality of the prompt you write.

At TaskWork, we see teams transform their productivity every day when they learn to communicate effectively with the features of Asana AI, whether through smart rules, project creation with AI Studio or AI Teammates.

In this article, we explore why specialized prompts matter, how Asana AI “thinks” and a practical method to start getting the most from these tools.

Why is the prompt the real “engine” of AI in Asana?

Think of a prompt as a project brief. If you simply say “make a plan”, the result will be generic. If you specify the objective, audience, deadline and available resources, the deliverable will be much more precise.

The same applies to AI. As Asana AI Program Manager Ethan DeWaal explains: “Think of AI as a golden retriever: it wants to please you, but you need to be specific about what you want.” Asana AI, whether an AI Teammate or a rule created in AI Studio, navigates the Work Graph, the structure connecting tasks, projects, goals and portfolios, to understand your work’s context. However, it can only do this effectively if you provide the right instructions.

Here are the main reasons to invest time in building good prompts:

Specificity creates relevance: A vague prompt produces generic answers. A well-structured prompt with clear context and goals produces results you can use immediately or with minor adjustments.
AI does not “guess”: It does not know what you do not tell it. Unless you specify that one report is for executives and another for the operations team, AI cannot adjust the tone or level of detail.
Less time and fewer iterations: A good first prompt drastically reduces back-and-forth. As the Asana Marketing Ops team’s example showed, manual reviews that took 3 to 4 hours were completed in 45 minutes with a well-configured AI Teammate.

How does Asana AI “understand” your request?

Asana offers several ways to interact with AI, each with its own characteristics:

AI Studio (Smart Workflows): Here, you describe the automation you want to create in natural language. The “Words to Workflows” feature interprets your description to build triggers, conditions and actions, and even suggest AI variables. For example, instead of manually configuring a complex rule, you can simply describe it: “When a high-priority task is created, assign it to João and set the due date to 3 days from now.”
AI Teammates: These are collaborative AI agents with persistent “memory”. They understand your projects’ context and can be triggered by mentions, rules or assignments. To make them act as real assistants, you define their “Behavior Guidance” (how they speak, think and deliver work) and provide clear examples of what you expect.

In every case, Asana AI benefits from a structured approach that helps it interpret your instructions correctly. This is where a powerful framework for communicating with AI comes in.

The four-pillar method for effective prompts

To help you create effective prompts without unnecessary complexity, Ethan DeWaal developed a simple framework that can apply to any situation, whether creating a project, building a rule or asking an AI Teammate for help.

Use the acronym PMTC (from Portuguese: Persona, Meta, Tarefa, Contexto — Persona, Goal, Task, Context) to remember the four essential components:

Persona: Define who the AI should “be” for this task.	Example: “You are an expert in marketing planning and project management.”
Goal: What overall objective do you want to achieve?	Example: “Your goal is to help me structure a clear, actionable product launch plan.”
Task: What specific action do you want the AI to perform?	Example: “Your task is to create an Asana project based on these meeting notes.”
Context: Provide all necessary details, constraints and examples.	Example: “Use the ‘Product Launch’ project template in our portfolio. The final deadline is November 30. The budget is BRL 50,000. Here are the meeting minutes with the initial decisions.”

Putting it into practice

To create a project in AI Studio: Instead of “Create a marketing project”, use the PMTC framework: “You are a project manager (Persona). Your goal is to create a project for a digital marketing campaign (Goal). Your task is to generate a project with Content Creation, Distribution and Results Analysis sections (Task). The project must be completed in 4 weeks, with a BRL 10,000 budget. Base it on last quarter’s campaign template in Google Drive (Context).”
To delegate to an AI Teammate: Instead of “Review the project status”, you could ask: “Act as an operations analyst (Persona). Your goal is to give me a clear picture of the ‘New Website’ project’s progress (Goal). Your task is to analyze the project and flag overdue tasks, blockers and risks (Task). Prioritize tasks that affect the final delivery date and provide an executive summary (Context).”

Refinement is the key

The first version of a prompt is rarely the final one. The secret is iteration. As DeWaal suggests, if AI did not deliver what you expected, give direct feedback: “Make the tone less formal”, “Be more concise” or “You left out cost data”.

Imagine training a new intern: you give instructions, review the result and provide feedback for the next attempt. AI follows the same logic. The advantage is that, over time, AI Teammates build memory and learn your preferences, reducing rework.

Conclusion: the future of project management depends on clear communication

AI in Asana is powerful, but its full potential emerges when we give it well-crafted prompts. Whether automating repetitive tasks, creating complex projects from scratch or gaining strategic insights, the quality of your communication with AI will determine the quality of the results.

Start applying the PMTC framework to your next requests. Provide context, be specific and refine as needed. With a little practice, you can have a virtual assistant capable of transforming the way your team works.