The hype around AI and deep learning caught pur attention when the first breakthroughs started to happen and true image recognition and classification suddenly became a real thing. The long lasting AI winter started to end quickly.

Google released TensorFlow, an open-source machine learning library and suddenly everyone could start to run its own experiments at home. As we delved deeper into its capabilities, we knew we had to learn more about it. I even enrolled in the Deep Learning specialization on Coursera offered by deeplearning.ai, which covered topics such as neural networks, convolutional neural networks, and recurrent neural networks and got my first certificate in 2017.

After the first steps in this world of AI, a question quickly came up: How can we really put this tech into something productive and not just gimmicky?

So, at Magnolia, we went on using DeepLearning4J to develop our first recommender system for authors and shipped our own image classification network based on ResNet3. However, it became clear quite fast that the actual development of state-of-the-art solutions was progressing very fast and keeping at the same level in-house was impossible given the massive research budgets of specialized companies. What also became clear was that AI was still in a research phase with no real production-ready services that could help archive real outcomes and value.

That was starting to come two years later when image recognition was suddenly available as a service from multiple vendors at very competitive rates. Having automated image tagging helped already quite a bit to boost editor productivity on image lookup.

The first service that really helped to raise productivity came with translation.

Translation services from Google and Microsoft existed for a long while and they produced results that somehow helped to understand the intent of the original text. However the overall quality was not ready for real world delivery. That changed when DeepL entered the scene. And again, soon there was a service available that could be integrated quickly in our translation processes.

DXplained Webinar: Making generative AI real for content teams

Learn how to take advantage of AI to save time and energy for marketers and other editors throughout the content lifecycle, while giving them control with a powerful workflow.

Translation

As businesses increasingly operate on a global scale, the ability to provide content in multiple languages is becoming more important. Magnolia has integrated with DeepL, an AI-powered translation service that delivers fast and accurate translations. DeepL uses neural networks to understand the context of a sentence and translate it accurately. The integration with Magnolia means that businesses can easily provide content in multiple langua ges, without the need for human translators.

Content classification

Content classification is an essential aspect of content management. Magnolia leverages AI and ML to classify content efficiently and accurately. Magnolia uses natural language processing (NLP) to analyze the text of a document and automatically assign it to the appropriate category. This saves time and ensures that content is consistently classified, making it easier for users to find what they are looking for.

Image generation

Visual content is an essential part of digital experiences, but creating high-quality images can be time-consuming and expensive. Magnolia has integrated with DALL-E 3, an AI-powered image generation service developed by OpenAI. DALL-E 3 uses deep learning to generate images from textual descriptions. For example, you can describe an image in words, and DALL-E 3 will create a high-quality image based on that description. This integration allows businesses to quickly and easily create images for their content, offering creative freedom and making stock images, photoshoots, and manual design optional.

Image recognition and tagging

To make it easy for you to find matching images when creating a page, AI automation can analyze images. Magnolia’s image recognition feature automatically tags image assets, making it easy to search for specific images among your assets. Extract insights, and add descriptive tags using image classification, so that you can easily search for them using keywords. You can use a local, custom-built neural network, or integrate Amazon Rekognition for advanced image recognition.

Text generation and automatic content optimization

Magnolia has also integrated with Chat-GPT, an AI-powered text generation service developed by OpenAI. Chat-GPT uses deep learning to generate text that is almost indistinguishable from human-written text. This integration allows businesses to quickly and easily generate content for their websites and digital experiences. For example, you can describe a product or service in a few words, and Chat-GPT will generate a detailed description that can be used on a product page.

You can use this feature in Magnolia to automatically generate components, component variants for personalization, stories, or even entire pages, including OG and SEO relevant metadata.

So, in a nutshell, Magnolia is now fully supporting AI to enhance content management throughout its entire lifecycle. And we’ve bundled all these features into our AI Accelerator.

With these capabilities, Magnolia’s AI Accelerator gives you a tremendous edge in productivity, the building blocks to get you started with every piece of content you can think of, and that you can further tweak, improve, or add your brand’s own personal flair. With Magnolia’s powerful workflows, content teams stay in control, ensuring that any content is fully reviewed and on-message before it’s seen by the audience it’s intended to reach.

We use proven AI services such as OpenAI ChatGPT, DALE-E, Amazon Rekognition, Amazon Comprehend and DeepL. With our composable approach, you can freely use other services, including your own custom developed, to match your company’s AI strategy and governance needs, and always stay on top of new technologies.

All services are available in one unified UI in Magnolia, comfortably embedded in the workflow that content editors already know and love. No context switching between tools, no copy/paste, no silos.

If you’d like to learn more on how you can apply AI to improve your content lifecycle and get a peek into Magnolia’s AI Accelerator, watch their webinar recording from October 25th.