MediaFlows, Cloudinary’s workflow automation engine, automates visual media tasks. From workflows that occur within Cloudinary, to workflows that involve connecting to other platforms in your visual media ecosystem, Cloudinary’s MediaFlows is purpose-built and integrates natively with Cloudinary’s asset pipeline.
In this post, we’ll examine how you can apply workflow automation using MediaFlows to moderate user-generated content (UGC).
UGC can be a challenge in the visual media space. You want your customers, partners, users, or the public to upload their own content into your app or website, but also need to ensure the content is relevant, compliant, and on-brand. For example, in marketplaces, inappropriate content can lose customers’ trust or dilute your brand. Like seeing poor-quality images of food or a book summary on a review site for restaurants.
That’s why you have to moderate content. But doing so efficiently and effectively is another challenge. Moderating every single uploaded image or video is time-consuming and resource-intensive.
Workflow automation like Cloudinary’s MediaFlows can solve this problem.
In this case, we’ll use MediaFlows’s PowerFlows, the low-code, flexible option for you to take full control of your workflows.
Let’s begin in Cloudinary Assets, Cloudinary’s AI-powered digital asset management (DAM) platform. This is where Cloudinary displays the assets that exist in your Media Library. The workflow that we’re creating will be created inside the Cloudinary console, too. Our goal is to automate decision-making so that only relevant and brand-safe content gets published.
Let’s examine this preprepared UGC moderation workflow:
As with any workflow automation, we’ll need to begin with a trigger. In this case, it’s an upload to Cloudinary, as you can see in the first block: Cloudinary Upload.
The blocks that come afterwards determine the actions. In this case, we’ll apply the “two shoes test” using Cloudinary AI Vision to determine whether the uploaded image contains two soccer shoes (if there’s only one soccer shoe or none, the image will move to manual moderation). The next step is a JQ processor that checks the answers coming from the AI Vision moderation. If the answer is “yes” the asset is approved and added to the site/app. If it isn’t, we can send the asset to a manual moderation process, which will be done in Cloudinary.
Using MediaFlows’ logs, you can track which assets went through which path, so you can check each moderated image to see whether the workflow is doing what it should do.
Moderation workflows are more than just using AI to determine whether a user-generated image or video is compliant or not.
You can use the same workflow and add more actions. For instance, if an asset passed moderation, you can add tags or alt text to the asset, such as “soccer shoes” or anything else that fits the workflow.
You can also add a step to notify moderators that an asset entered manual moderation. You can create an email or send the task to a task management system, integrate the task into Slack, and more. For instance, you can trigger a Slack alert to a #moderation channel when an asset fails AI screening.
You can also use MediaFlows’ http block and really integrate it with anything. MediaFlows gives you the freedom to send whatever data you want to whatever endpoint you’re using. So, if you’re using an internal service within your company that has an API and you want to communicate with it, you can simply use the http block.
This approach to moderation can be extended to other workflows, such as campaign asset QA, contributor uploads, or even branded hashtag challenges where moderation speed and scale are critical.
Want to try MediaFlows for yourself? Sign up for a free Cloudinary account today to get started.
