Major

Asana Creative Request Intake Agent: End-to-End Guide

Learn how to build an Asana creative request intake agent that verifies briefs, communicates with requesters, and creates accurate tasks.

Ritvij Gautam
software engineers planning sprint board (Asana AI agents and automations)

An Asana creative request intake agent helps teams manage incoming creative and marketing work by verifying briefs, following up on missing details, and generating complete Asana tasks. This process improves clarity and saves time for both requesters and creative teams. In this guide, you will learn how to build and operate an effective intake agent using Asana, Gmail, Slack, and Google Drive.

This guide covers one Asana process end to end. For the Asana API guide, connection setup and the full list of Asana use cases, see More Asana use cases.

Why This Process Breaks Today

Creative and marketing teams often juggle high demand for design, collateral, and campaign assets. Requesters submit work requests through forms or emails, but the details in these requests are usually incomplete or unclear. This triggers multiple back-and-forth messages: missing drive links, ambiguous objectives, or incomplete specifications force project managers to chase down details before any real work can start.

Teams base projects in Asana, but when intake processes are manual, poorly-formed tasks pile up. Without a system for context gathering and validation, creative teams waste time clarifying requests instead of delivering work. Balls get dropped; the wrong files are delivered; due dates slip due to missing information.

The Systems and Data Involved

An Asana creative request intake agent must move across several tools to gather, verify, and submit the right information. The primary systems and data objects involved include:

Asana

  • Projects: The containers for different types of creative work
  • Tasks: Individual work items, usually created via the POST /tasks endpoint
  • Custom fields: Capturing details such as "Campaign Name" or "File Type"
  • Task attachments: Reference files or images

Form Intake

  • Source varies: Google Forms, Typeform, or internal web forms. Submissions typically arrive in a spreadsheet, via webhook, or an email inbox.
  • Typical fields: Requester contact, project name, description, due date, drive link, asset type, campaign goals

Gmail

  • For sending follow-up emails if required fields are missing
  • Endpoint: send via Gmail API

Slack

  • Optionally for instant clarification, if requesters are on Slack
  • Endpoint: chat.postMessage

Google Drive

  • Where creative briefs or asset references are stored
  • Endpoint: files.get for file verification

What is Code vs. Agent Judgment vs. Human Approval

Deterministic Code

  • Receiving form submissions and parsing the data
  • Validating basic field presence: is there a due date, is the drive link present, are required custom fields filled?
  • Looking up or mapping the correct Asana project for the request type
  • Creating an Asana task with well-formed fields

Agent Judgment

  • Assessing if the brief is truly complete versus superficially filled (e.g., vague objectives or unclear specs)
  • Interpreting answers to open-ended questions ("What is the main message?")
  • Estimating effort level based on description and attached reference assets
  • Deciding the best communication channel for clarification (email vs. Slack)
  • Drafting clear, context-specific follow-up questions to the requester

Human Approval

  • Can involve final review for high-priority or ambiguous requests before creating the Asana task
  • Optionally required for requests above a certain estimated effort threshold

Worked Example: Walking One Record Through

Record:

  • Submitted via Google Form
  • Fields:
  • Name: Alice P.
  • Email: alice@example.com
  • Request Type: Web Banner
  • Campaign Name: Summer Sale
  • Due date: 2024-07-01
  • Brief: "Summer sale graphic for our homepage."
  • Google Drive link: empty
  • Slack handle: @alicep

Step 1: Intake

  • The workflow receives a new form response webhook
  • Code parses the submission

Step 2: Basic Validation

  • Checks for all required fields
  • Notices the Drive link is missing

Step 3: Agent Review

  • Determines that the brief is incomplete: while there is a campaign name and general description, there’s no sample/reference or clear direction on dimensions, messaging, or target audience.
  • Composes a Slack message: "Hi Alice, could you please provide a Google Drive link to any reference files for your Summer Sale banner? Also, can you specify the preferred dimensions and main message for the banner?"
  • Sends the message to @alicep on Slack

Step 4: Human Response

  • Alice replies in Slack, sharing a Drive link and clarifying, "Use homepage hero specs: 1600x600. Main message: 'Shop Summer Savings!'"

Step 5: Agent Finalization

  • Verifies the Drive link points to an accessible folder with images
  • Estimates this as an "Intermediate" workload based on asset complexity
  • Drafts the Asana task with these details

Step 6: Task Creation

  • Code creates a new task in the "Marketing Requests" Asana project using POST /tasks with fields:
  • Name: Summer Sale Homepage Banner
  • Notes: includes full brief, Slack thread link, and Drive link
  • Custom fields: Asset type, Campaign name, Workload estimate
  • Due date: 2024-07-01
  • Follower: Alice

Building It on Major

The App

Define an app in Major that:

  • Connects to Asana using scoped credentials, restricting writes to the marketing project
  • Connects to the email inbox receiving intake forms
  • Connects to Slack with the ability to message requesters
  • Connects to the Google Drive folder where assets are stored

The Agent

  • Reads parsed form submissions
  • Calls code to check for missing fields
  • Reviews open-ended brief fields for substance; flags entries that are vague (e.g., "need a graphic")
  • Decides what clarifying details to request and how to phrase the follow-up
  • Decides effort estimate ("basic," "intermediate," or "advanced") based on brief and assets
  • Documents all facts, decisions, and outbound messages in the task creation log

The Workflow

  • Trigger: New form response (webhook or new email)
  • Step 1: Parse and validate submission (app, code)
  • Step 2: Agent checks completeness
  • Step 3: If incomplete, agent drafts questions and sends via Slack or Gmail (app + agent, based on contact info)
  • Step 4: Wait for user response, then re-validate
  • Step 5: If complete and under effort threshold, app creates Asana task
  • Step 6: (Optional) If request is very large or complex, route for human review before creating the Asana task
  • Step 7: Log all actions in Major for traceability

What to Measure

  • Time to first response: From intake to initial follow-up with the requester
  • Time to clear brief: From intake to all required information gathered
  • First time completeness rate: Percentage of requests that are complete at submission
  • Task creation accuracy: Number of Asana tasks with missing or incorrect fields
  • Requester satisfaction: Optionally solicited via follow-up email

Guardrails (What the Agent Must Never Do)

  • Never create Asana tasks with missing required fields or ambiguous brief information
  • Never guess or fabricate missing details
  • Never send sensitive data or project links to unlisted email addresses or unknown Slack users
  • Never bypass human approval for high-effort or out-of-scope requests
  • Never operate outside of designated Asana projects or Google Drive folders

Frequently asked questions

Does the intake agent handle file uploads to Asana tasks?

The agent verifies that Drive links are accessible and correct but does not upload files directly to Asana by default. Instead, it includes Drive links in the task for the creative team.

How does the agent decide if a brief is complete?

The agent uses a checklist: required fields, presence of reference materials or dimensions, and clarity of the stated objective. If any are missing or vague, it follows up with the requester for clarification.

What if a requester does not respond after follow-up?

The agent can be configured to send a reminder after a set interval. If there is still no response, the request is held and not entered into Asana until all key details are provided.

Can the agent handle approvals for larger requests?

Yes. If the agent estimates that a request exceeds a defined effort threshold, it routes the request to a human reviewer for manual approval before the Asana task is created.

Start building on Major

Connect Asana, describe the process, and Major builds the app that holds its state, the agent that makes the judgment call and the workflow that runs it, with scoped access and an audit log from the first deploy.

Frequently asked questions

What are Asana AI agents and how do they connect with Asana?
Asana AI agents are automated agents built with Major that use Asana’s REST API to interact with tasks, projects, and workspaces. These agents can monitor, analyze, or take action on Asana data without manual effort.
How does Major automation differ from Asana’s own automation features?
Asana's automation handles repeatable actions inside the platform, like task assignments or status updates. Major enables cross-tool automation by orchestrating actions between Asana and other systems, including approvals or escalations.
Can agents make decisions or change tasks without human intervention?
Agents can perform routine updates, monitor progress, or flag issues but decisions involving judgment, like escalations or critical changes, always require a person’s approval within the workflow.
What authentication is required for agents to access Asana data?
Agents must authenticate with either OAuth 2.0 or a Personal Access Token, provided via the Authorization header. An Asana account is required, and OAuth requires registered apps.
What is the rate limit for API calls with Asana agents?
Free domains are allowed up to 150 requests per minute per token, while paid domains have a limit of 1,500 requests per minute. There is also a 60 requests per minute limit for Search endpoints.