WORKFLOWS / PROJECTS AND CHANGE
Turn a vague request into a workable project.
Work out what a request needs, check whether related work already exists, and agree on a manageable scope before creating tasks.
WHAT THIS WORKFLOW PRODUCES
A project scope people can agree to.
A short record of the result, the work involved, and the decisions still needed. Straightforward requests can stay as individual tasks. A project is useful when the work needs several steps or people. The example below is a proposed scope: owners and dates become commitments only after agreement.
CLIENT HANDOFF IMPROVEMENT
ILLUSTRATIVE DRAFT · AWAITING APPROVAL
THE OPERATING METHOD
Understand the workflow without AI.
Start with the request as it arrived. Clarify the result, investigate the existing work, and agree on the smallest useful piece of work. AI is introduced only after the method makes sense on its own.
Clarify the result
What should be different when this is finished?
YOUCheck existing work
Find the relevant process, project, or previous attempt.
YOUSet the boundaries
Agree on what is included, excluded, and still unknown.
YOUConfirm responsibility
Identify the person doing the work and the person accepting the result.
YOUCreate the work
Add only the tasks needed for the approved scope.
YOUA clear request can go straight to an existing task. Use the full sequence when the request needs interpretation or coordination.
WHERE THE FACTS LIVE
Use the systems where the work already lives.
The tools below are examples. Use the team's existing request channel and project system where possible. Access to a conversation or document must be authorized by its owner. Named tools link to their pages in the library.
THE ORIGINAL REQUEST
Requested outcome, deadline, and supporting explanation.
WORKING REFERENCE
Current procedure, agreed requirements, and relevant documents.
THREE WAYS TO RUN IT
Choose how much machine.
Choose the version that matches how requests arrive. The third version is useful for repeated intake. Occasional requests usually do not need an automated intake system. The tools named are examples. Use what your organization already runs, provided the connection is supported and approved.
LEVEL 1
Clarify the request in chat.
Give AI the request and the facts you already know. Have it identify the few unknowns that would materially change the work, then draft the scope after you answer.
Use this version when a request is vague, politically sensitive, or difficult to turn into a first step.
WHAT THIS VERSION REDUCES
Rewriting the request and overlooking basic scoping questions.
WHAT A PERSON STILL DOES
Whether the work is worth doing and what the requester meant.
MAIN LIMITATION
AI cannot know existing commitments or internal history that you have not provided.
Build this version, step by step 4 steps, one action each. Prompts carry a Copy button.
Paste the request exactly as it arrived.
ORIGINAL WORDING
Include the request itself, any fixed deadline, and the team or process it concerns. Take out anything your AI account is not approved to handle.
Ask for the questions that would change the answer.
Help me scope this request. Ask up to three questions whose answers would change the outcome, scope, or first step. Do not invent owners, deadlines, budgets, or requirements. First tell me if this could be handled as a single task.
Answer the questions, then ask for a draft scope.
ONE SHORT RECORD
The scope says what the result should be, what is included, what is excluded, the first few steps, and the decisions still open. Anything unknown stays visibly unknown rather than being guessed.
Agree the scope with the person who asked, before anyone is assigned.
THE AGREEMENT
Go through the scope with the requester. Copy the agreed version into the project tool the team already uses, and confirm the task owners with them directly.
LEVEL 2
Scope against existing work.
Connect an AI assistant to the approved project system and reference documents. It looks for related work, identifies useful material to reuse, and drafts a scope with links to its findings.
Use this version when new requests often overlap with projects, processes, or previous attempts.
WHAT THIS VERSION REDUCES
Searching for related work and rebuilding a scope from scratch.
WHAT A PERSON STILL DOES
Whether the existing work really covers the request and whether people can take on more.
MAIN LIMITATION
An empty search does not establish that no related work exists. Permissions, naming, and incomplete records affect the result.
NOT READ-ONLY BY DEFAULT
Asana's MCP tools operate within the connected user's access. Do not assume this connection is read-only. Review the tools and approval controls available in your AI client before connecting it.
Build this version, step by step 5 steps, one action each. Prompts carry a Copy button.
Connect the AI assistant to the project tool.
CLAUDE AND ASANA
Asana's official MCP connection gives an AI assistant a set of approved ways to read and act in Asana. Add Google Drive only if the reference documents live there.
Tell it where to look, and that it may only read.
ONE TEAM OR PROJECT
Name the projects and documents it may search. Ask for research and a draft only, with no tasks created or changed.
Ask what already exists.
REUSE, EXTEND, OR NEW
Ask the AI to sort what it finds into work that can be reused, work that needs extending, and work that is new. Every suggestion links to a source or says that no source was found.
Check the gaps with the people involved.
THE CHECK
- The older work it found is still valid.
- The affected owner has said whether they have capacity, rather than an empty task list standing in for an answer.
Approve the scope, then create the tasks.
APPROVED SCOPE ONLY
Read the final scope, then create or amend the tasks. If the AI does this through the connection, approve the exact changes first.
CAPABILITIES CHECKED AGAINST OFFICIAL DOCUMENTATION
Asana's MCP tools reference documents project and task retrieval, search, and supported actions; some search tools require a paid Asana plan. Source, checked September 6, 2026.
Asana documents an official MCP connection for compatible AI clients, with its own connection and authorization setup. Source, checked September 6, 2026.
LEVEL 3
Give incoming requests a scoping workflow.
A request submitted through the team's own Asana form starts a smart workflow inside Asana. It checks for duplicates, searches the named projects and the connected Drive folder for related work, and writes the draft scope and its open questions onto the task. The intake owner approves; a rule creates the agreed tasks. Nothing to host, no second tool.
Use this version when the team receives enough similar requests that repeated scoping is creating a backlog.
WHAT THIS VERSION REDUCES
Repeated intake triage, the search for related work, and first-draft scoping.
WHAT A PERSON STILL DOES
Priority, ownership, resource allocation, and acceptance of the work.
MAIN LIMITATION
A smart workflow will not understand your intake standards simply because it runs in your project tool. It needs one good scope as its example and permission to stop when the evidence is thin.
WHAT THIS NEEDS, AND THE ALTERNATIVES
Asana Forms and AI Studio come with the paid plans, with a monthly allowance of AI runs. Monday.com and ClickUp carry forms and automations of their own. Where intake spans apps, a form in Tally feeding a CRM for example, Make or n8n coordinate it, with an agent for the research step.
Build this version, step by step 6 steps, one action each. Prompts carry a Copy button.
Put all requests through one Asana form.
THE FORM
The form asks for the request, the result the person wants, the team involved, and the deadline if there is one. Every submission becomes a task in the Intake project with its own ID. If the team already has an intake form, keep it and connect the workflow to that.
Add an AI Studio smart workflow that runs on each submission.
FOUR ACTIONS
It checks whether this duplicates an open request. It searches the named projects and the approved Drive folder for related work. It writes a draft scope into the task description: the result, what is in, what is out, the first steps, and the open questions. Then it assigns the task to the intake owner.
Show it one good scope.
THE MODEL
Paste one approved scope into the workflow as the example to copy. Tell it to write "Needs answer" and stop, rather than guess, whenever the request does not say enough.
Close the loop with the person who asked.
HOW AN ANSWER COMES BACK
When a task is marked "Needs answer", the intake owner sends the question to the requester in their own name. The reply goes on the task as a comment, the owner changes the status to "Ready", and that change runs the workflow again with the answer included.
Let approval control what happens next.
ONE FIELD, ONE RULE
Add a field called "Scope approved" that only the intake owner can set. Add a rule: when it becomes Yes, move the task to the project and create the agreed subtasks. If the scope changes after approval, the field resets.
Test it before you trust it.
THE TEST SAMPLE
- A request with half the fields empty.
- A duplicate of an open request.
- A request that overlaps existing work.
No task should be created without the field, and submitting the same form twice should not create two sets of tasks.
CAPABILITIES CHECKED AGAINST OFFICIAL DOCUMENTATION
Asana AI Studio documents smart workflows triggered by form submissions, with Check, Research, Create, and Route actions and human approval checkpoints. Source, checked September 6, 2026.
Asana AI Studio documents human approval checkpoints inside a workflow. Source, checked September 6, 2026.
Asana documents Forms on the Starter plan and above. Source, checked September 6, 2026.
Asana's pricing page lists AI Studio with a monthly allowance on its paid plans. Source, checked September 6, 2026.
n8n documents a tool-using agent node as an alternative for intake that spans apps. Source, checked September 6, 2026.
WORKS EVEN BETTER WITH
Once the scope is clear, check the risks, agree on responsibility, and decide where the work will be tracked.
Project Premortem → | Kanban → | RACI and DACI chart builder →