header-mobile-bg

Utilizing AI to meet the demand for personalized content and content optimization

Oliver_Heyden_Photo-2 Colorized_4

Oliver Heyden

Soon after OpenAI released ChatGPT to the public, the hype began. It seemed like everyone was talking about AI, particularly in the realm of natural language understanding and generation, and its possibility to revolutionize the marketing world.

However, alongside its many fascinating possibilities, people have questioned its real-life value for businesses. Is ChatGPT really as good and revolutionary as it seems to be? Or are some key features still missing?

If you try out the standard services on your own, you’ll quickly notice their limitations. Many have complained that ChatGPT cannot answer every question in the world, and that its knowledge was outdated.

To experts in this field, like the tech company wonki, based in Germany, this does not come as a surprise. The language models ChatGPT or GPT-3 are great for a variety of natural language tasks, like text summarization, answering questions and translation. However, they are neither meant to provide the most accurate answer, nor make decisions for you. Rather, they are simply outstanding tools for understanding and generating natural language.


An extended approach for businesses

As I wrote in last week’s blog post “Leveraging ChatGPT to drive impact with CoreMedia Content Cloud”, there are many use cases to support businesses with the quick and easy generation of texts. The outstanding questions, though, are: how good is the quality of texts that are generated by the AI? What facts is text generation based on? And can AI really consider my company’s specific language and my business domain? All these questions are not answered by the pure GPT or ChatGPT language models.

And this is where CoreMedia’s Partner wonki comes into play. wonki specializes in providing content services based on GPT-3 that is here to solve the “content crisis” – the massive need from businesses to create content and content variations to fulfill the demands of their individual audiences.

wonki offers extended services, such as training the language model with the specific language domain of a business. They also have answers on the topic of privacy, as they can rely on GPT-3 services that are not-US based but rather European.

Those were the reasons why we at CoreMedia were keen to integrate with their services into our CoreMedia Content Cloud.


Integrating AI content services into CoreMedia Content Cloud

Thanks to CoreMedia Content Cloud’s composable nature, wonki’s integration was pretty straightforward!

The first service we integrated was wonki’s GhostwritR. We completed the integration via our CoreMedia Feedback Hub. In the following screenshot, you can see the user interface for entering an idea – and then GhostwritR subsequently generates the desired text.

GhostwritR

GhostwritR

You will notice a key difference to a standard ChatGPT or GPT services — GhostwritR also returns a confidence score, telling you how valid the answer is. This is not just a quantitative number. You also understand its reasoning by seeing the sources for the AI answer. This is shown in the next screenshot:

For a business user, this will make a huge difference, as you now have concrete hints as to what is happening in background of the AI.

There are also other expert services available in wonki’s offering that we have just integrated as well.

One important service is the SummarizR. This service creates summary texts or excerpts. It also allows you to adjust a text to generate a different tonality or a different wording that is suitable to specific target consumer audiences. Or it can rephrase an existing text, which is important if you think about optimizing your content freshness – this is highly relevant for both your site visitors, as well as for your search engine ranking (fresh, frequently updated content gives a higher search ranking.) The following screenshot shows the SummarizR in action:

And finally, there is the TransformR. This is used for several use cases:

  • To extract keywords or tags from your text
  • To generate catchy taglines
  • To generate HTML metadata or SERP texts (as shown in every Google search result)

All of this is used to optimize metadata, mainly for SEO. See the TransformR in action within CoreMedia Content Cloud in the following screenshot:

TransformR

TransformR



Personalized content in action — all thanks to AI  

You can see that there is a huge difference if you’re just playing around with some (rather fascinating!) standard AI services, such as ChatGPT (or DALL-E) – but it’s a different story entirely as to how you can really benefit from these services in a business environment, with its own language domain and where data privacy is a topic.

The AI-driven content services from wonki, based on GPT, tackle these challenges. In combination with our digital experience platform CoreMedia Content Cloud, the pieces of the puzzle come together.

Business can now leverage these services to create content in high quality at a more massive scale than ever. The demand for content increases every day, and businesses’ goals to target their audiences more specifically in terms of the right tone of language (targeting “Generation Z” vs the “Babyboomers” is certainly different) or the way to explain your products and services (more factual versus more imaginative, or in more detail versus in all brevity) is just ever growing – essentially, THE foundation necessary to create scaling personalized content experiences.

If you are interested in discussing this further, please get in touch with my colleagues or me — there is so much more to come!