Exploring the ResourceSpace-OpenAI integration

10th October 2024

Artificial intelligence and machine learning technology is impacting every sector in the economy, and this is particularly true for software such as Digital Asset Management.

We’ve discussed in length how ResourceSpace is leveraging AI previously, but in this article we’re going to take a closer look at a real-life use case of the ResourceSpace-OpenAI integration.

The challenge

MCC (also known as the Menonite Central Committee) is a Christian-based relief agency working in development and peace in more than 40 countries.

The agency relies on ResourceSpace to organise and manage the many photos acquired by their offices across the world. MCC also commissions freelancers to go on content gathering trips to help support the agency’s storytelling efforts. However, all of this content presented some challenges to MCC, particularly when the International Press Telecommunications Council (IPTC) added alt text to their photo metadata standards for news media.

READ MORE: The Complete Guide to Metadata

The IPTC Photo Metadata Standard is the most widely used standard for describing photos, and it provides the structure and definition for metadata properties. Although there were some metadata properties embedded in photographs that ResourceSpace could copy to fields using the EXIFtool—such as the type of camera used and date the picture was taken—the alt text for the image would have to be generated manually by the editors.

This was an issue for two reasons:

1. Creating alt text for every image being acquired by MCC would take a lot of time
2. The editors uploading the images to ResourceSpace aren’t trained in content production and some aren’t native English speakers.

The function alt text serves is also nuanced. For web users who are visually impaired, alt text can be read aloud by Screen Readers, meaning it should describe what the image depicts rather than providing narrative. For example, the alt text for the featured image of this blog is: ‘A person is crouching outdoors, collecting water from a pipe into a bucket, surrounded by greenery, possibly in a rural or agricultural setting.’

By contrast, a Description field for this image might provide additional context about who the person is and the work they’re doing. In this example, the Description provides this information:

“Jayram Pradhan is one of the most innovative and progressive farmers in his village. He uses small pipes to extend the community pipe network throughout his farm to irrigate his crops and vegetables. Since 2014, this MCC-supported water project has been nourishing the land and enhancing livelihoods and health in Sinisingi village communities. Pradhan grows a variety of crops and increases his income with ease.”

While that’s interesting information, it doesn’t serve the purpose of alt text and should be in a separate field.

How the ResourceSpace-OpenAI integrations solved it

ResourceSpace’s OpenAI integration allowed MCC to offload the task of creating all of that alt text. Now when images are uploaded to the DAM, the GPT language model looks at the photo and describes it in 250 characters or fewer (the IPTC standard) automatically.

This doesn’t just ensure MCC is compliant with the IPTC standards, but it also saves the image editors untold hours they’d have otherwise had to spend writing the alt text themselves.

Below is another example of an image uploaded to MCC’s DAM:

The alt text that was automatically generated for this image was ‘A person stands in a field with rows of crops, holding a red basin. The sun is low, casting a warm light over the scene with trees and a hill in the background.’

That might not seem overly complex, after all it’s just describing what’s in the image, but ensuring it’s within 250 characters, accurately describing the image and writing in proper English can be tricky. Just three minutes spent writing alt text would take five hours for a collection of 100 images.

Setting up the OpenAI integration in ResourceSpace was incredibly easy and was simply a case of activating the plugin and writing the prompt, but now it’s up-and-running it’s generating automated alt text for every image submitted into MCC’s DAM system.

If you’d like to see for yourself how ResourceSpace’s OpenAI integration can streamline and speed up your Digital Asset Management workflows, book your free demo of the system below. Alternatively, launch your free ResourceSpace instance immediately and see what the platform has to offer.