Getting Started with AI Workflows
AI Workflows is intended for staff who are able to design and maintain their own workflows. ITS provides platform stewardship, documentation, templates, and limited consultation to help users get started and build sustainably.
You do not need to code to get started with AI Workflows, but more advanced use requires some technical familiarity. Users should be comfortable thinking step-by-step and, over time, working with APIs, JSON, and data.To enroll in this service, you are required to complete the AI Workflows Training Canvas course.
Users are responsible for ensuring that workflows, integrations, and connected services align with the university’s data protection requirements. Always consult the U-M Sensitive Data Guide and Safe Computing guidance before building workflows.
Prerequisite Training
To gain initial access to the service, all users must complete the AI Workflows Training Canvas course (approx. 30 min). Once you complete the course and your request is processed, you will gain access to the AI Workflows environment.
In this course, you will learn:
- Core technical foundations
- Data structures such as JSON
- API literacy
- Expressions
- Loops
- Essential AI Workflows terms such as: nodes, triggers, webhooks, etc.
- Safe Computing and the Shared Responsibility Service Model
Logging in to AI Workflows
You must be connected to the U-M campus network (on campus) or to the UMVPN - All Traffic network (remotely).
- Open AI Workflows.
- Click Continue with SSO.
- If prompted, log in with your uniqname and UMICH password and complete multi-factor authentication (MFA).
Personal vs Project Workflows
AI Workflows supports two types of workflows:
- Personal Workflows (Personal Productivity)
- Created by default upon signup.
- Intended for personal automations benefiting only the individual user.
- Owned and managed by one user but can be shared.
- Limited to 250 scheduled executions per month with no option to increase.
- Project Workflows (Departmental/Team Use)
- Designed for team or department business processes.
- Accessible to all users associated with the project.
- Available upon request by submitting a form.
- Each workflow capped at 1000 scheduled executions per month; additional executions incur charges.
Note: A scheduled execution is any workflow run triggered by a time schedule, webhook, or any automated method other than manual execution.
After completing the Canvas course, you may begin creating your own personal workflows (for personal productivity), or you may request a project workflow (for departmental/team use).
Your First Workflow
Don’t start from scratch! Within 24 hours of first logging into AI Workflows, you will be granted "View" access to a dedicated U-M Templates Project directly within AI Workflows.
This project contains a collection of pre-configured workflows that represent common university use cases. You can browse these examples to see how they are built and import them into your own personal or team workspace.
Important Note on Credentials: To maintain security, templates do not include login credentials or API keys. After you import a template into your own workspace, you will need to "re-connect" the nodes to your own university credentials (e.g., your U-M Google account or Slack integration) before the workflow can run.
How to Use Templates:
- Browse the Project: Log in to the AI Workflows Template Project to view the library. Note: You must already have an active account and be on a campus network (or using the UMVPN - All Traffic) for this link to work.
- Follow the Guide: For step-by-step instructions on how to copy a template and set up your credentials, refer to our knowledge base article: How to Import and Customize Templates.