Reviews are one of the most important features in the intake and governance of AI use cases.
What this is:
The intention behind this feature is to provide a clear explanation of what the feature is and who it is for. By including a screenshot, users can visually see the feature in action, enhancing their understanding. This intro paragraph sets the stage for the rest of the article, guiding readers on what to expect and how they can benefit from the information provided.
What this isn't:
<optionally insert a diagram with a red x on it>
Potential misunderstandings for this feature could include confusing it with similar features or functionalities within the platform, misinterpreting the purpose or use case of the feature, or assuming it has capabilities beyond its intended scope. It is important to clarify any misconceptions and provide clear explanations to ensure users fully understand the feature and how it can benefit them.
Who this is for:
Here's how different users interact with Reviews:
Roles in Credo AI | Example Job Titles & Teams | Usage |
Governance Team (Admin) | AI Governance Team | Review intake forms, risk plans, and compliance |
Business User (User) | Director of Marketing Senior Engineer |
Create an intake form and request a review |
Product Owner (User) | AI Product Manager Sr Product Owner |
Create an intake form, assign contributors, request a review, and address concerns |
Data Scientist (User) | AI Engineer Data Scientist Data Engineer |
Invited as a contributor to an intake form, questionnaires, or for change requests in a Review |
How to use it:
Reviews start with an Intake Form from business users or AI development teams and end with Governance teams and other required approvals.
1. Business Users: Intake Form
Send business users a link to this page and instruct them to fill out a form
Get Started (New AI Use Case)
It's simple to get started, simply click Get Started or New AI Use Case and your intake journey has begun!
Choose a name
The first step is to choose a short, descriptive, unique name for your AI Use Case.
Fill our the Questionnaire
Next you'll be asked to fill out all the metadata and useful context about your use case, which will inform recommendations, alerts, and all go into your review.
Request a Review
When you're satisfied with your answers to your questionnaire and have answered them all, you'll be prompted to start a new review.
Add Reviewers for Feedback
You may automatically have a required review team added. You can also add specific teams and individuals to your review to get their sign-off and feedback.
2. Governance Admins: Conduct Review
Now that the user has requested their review, governance admins and reviewers will receive notifications to begin conducting the review.
Review Summary
The first tab of the review is the summarized most important information:
Critical Factors:
- Alerts: here you'll find the alerts that are based on rules your admins have setup when building the Questionnaire
- Questionnaire Alerts: Alerts triggered from answers to questions (specifically select boxes & radios)
- Risk Alerts: Alerts triggered from Standalone Controls for risk mitigation evidence
- Compliance Alerts: Alerts triggered from evidence in Compliance Controls in Policy Packs
- Governance Plan: the Risk Category (level), Risk Plan, and Compliance Plan will be populated here as the plan for governance of this use case.
- Risk Category: The level of risk from your chosen Risk Framework (for example, Minimal, Low, High, or Unacceptable from the EU AI Act Framework)
- Risk Plan: The list of Risks (Risk Type, Risk Scenario, and Inherent Risk Score) and their mitigation plan (via Controls)
- Compliance Plan: The required compliance and/or standards to be met via Policy Packs w/ Controls
- Feedback & Sign-off: the required and requested teams and individuals who are the stakeholders can leave their feedback here. The options are:
- Sign-off: Give the green light to move forward with the approved changes and governance plan
- Request Changes: Changes that you'd like done before the review is finished. If the review is approved with open Change Requests, they'll be noted for reference.
- Reject: Go on the record as a no-go vote for this review.
- Review Approval: This is the final decision for this particular review. Decide whether you want to:
- Approve the Review: Mainly for Low-Risk use cases, this is an approval with no requested changes or governance plan changes.
- Approve with Plan: This is the most common status, where the Review is approved to move forward with the agreed upon Governance Plan
- Cancel Review: This use case is not good to move forward at this moment.
Best practices:
Use a bullet list in order of the best ways to use these features
- Use it this way for these people with this intention
- You can always add this usage
- Don't forget you can do this thing
- Here's a stretch but would still work
What did we miss?
If there are missing features, features that feel wrong, or additional documentation or demo content that would help you use this feature in Credo AI, please request it here:
- Feature Request & Product Feedback
- Customer Support [Book Session] [Email Support]