How AI Is Transforming the CMS Industry

Discover the potential of AI when it comes to content creation, editing, and management, and see how it can take a CMS to the next level.

AI integration CMS blog header

Apostrophe's CTO, Tom Boutell, recently presented a talk at Philly Tech Week to share his experience using the OpenAI API to integrate AI into Apostrophe. If you are not familiar with this tool, Apostrophe is an open-source CMS built on modern technologies like Node.js and Vue.js that can operate as a traditional CMS, headless CMS, and website builder. 

To figure out whether integrating artificial intelligence was worth it, Tom first had to explore what new services like OpenAI have to offer. Digging into what AI is capable of doing, it becomes clear that it is the perfect assistant for editors and content creators to improve their productivity.

This is what inspired the development team behind Apostrophe to build a module that uses OpenAI APIs to bring the magic of AI into Apostrophe. Watch the recording of Tom’s talk for more details:

My Journey with the OpenAI API with Tom Boutell


To follow, you will get a summary of this journey, with insights into key points on whether artificial intelligence really is the future of CMS. Let’s jump into it! 

What Is AI?

AI, short for Artificial Intelligence, is a field of computer science that focuses on making machines able to perform tasks that would normally require human intelligence. It involves using data to develop and train models that enable computers to perceive, reason, learn, predict, and make decisions. 

As explained by Tom, AI can be defined as “computers doing work that we associate with brains.” The idea behind AI is to provide help or a better substitute for any task that used to require a human brain to be achieved.

It is important to point out that artificial intelligence is not something new. Its history dates back to the mid-20th century, with significant milestones such as the chess bot's victory over the world champion and the introduction of Siri. However, those systems were rather limited.

AI boom

In recent months, a new AI boom has emerged. The power of tools such as DALL-E, ChatGPT, and Bing Chat is impressive, sometimes even frightening. Compared to previous AI models, the new systems are trained with huge amounts of data, rely on complex learning techniques, and offer advanced predictive capabilities. Thanks to this progress, artificial intelligence has gone from being a cool feature to show to friends to an indispensable tool.

What Is AI Good For?

The ability of new AI systems to automate human tasks is shaping the future of the way we work. At the same time, there are specific industries, areas, and tasks where AI is more useful than others. Let’s take a look at some of those:

  • Automated content generation: AI can generate content based on given inputs or prompts. This includes blog posts, social media posts, product descriptions, images, and more. If content creation poses recurring challenges or you do not know where to start, AI language models can give you ideas, suggest improvements, and provide draft text. Copywriting has never been easier.
  • Avoiding repetitive tasks: AI-based tools can automate repetitive, boring, and time-consuming tasks. This allows us to enjoy more free time and focus on more strategic aspects of our work. For example, you can use artificial intelligence to summarize text, automatically format a document, or organize content into templates.
  • Content personalization and recommendation: Algorithms based on AI are able to analyze user preferences, behavior, and historical data to recommend content they might like. Online platforms use this feature to deliver tailored experiences, suggesting relevant articles, products, posts, and accounts to follow based on individual user interests.
  • Adapt existing quality content to new goals: Feed the AI models with quality work, and they will be able to generate new ideas, alternative versions, or adaptations suitable for different platforms or target audiences. This helps maximize the time invested in carrying out the original work. 
  • Improve the quality of your current work: Writing AI assistants are programmed to act as virtual editors, providing an additional layer of scrutiny to ensure the quality of your work. Think of Grammarly, for example. These tools are able to catch grammar and spelling errors, flag style inconsistencies, and suggest best practices based on your target audience. This also reduces the risk of oversights. Consider artificial intelligence a second pair of eyes.
AI benefits open source cms

CMS and AI: A Good Match?

What do all the tasks artificial intelligence is good at have in common? They are all, or nearly all, related to the world of content creation and editing. This makes AI the greatest ally of CMS (Content Management System) technologies. 

The reason they form such a great match is that artificial intelligence can integrate into content editing workflows easily and effectively. It does not take much effort to think about ways AI can support content editors and creators. Specifically, there are a lot of cutting-edge features that can be introduced to a CMS to enhance the overall user experience. 

To better understand the benefits that AI brings to the world of content management, let's look at how it can support the process of writing a blog post, from topic selection to promotion of the final article.

First, you need to find a convincing topic that is in line with your SEO and editorial goals, but also compelling enough to engage your audience. Running out of ideas? Ask ChatGPT to inspire you! Once you find the right topic and devise a good outline, you can get a first draft of the article you want to write in seconds. All you have to do is provide the AI model with the right prompts. Then, tools like DALL-E can help you generate visually-appealing images that align with written content. Writing a high-quality blog post becomes much faster thanks to these features.

Before hitting the publish button, you need to optimize the content for SEO and define a good meta description and title. Again, artificial intelligence can come to the rescue. By analyzing the text of your post, AI can figure out what it is about and give you the metadata you need to set to get the article ready for Google.

Next, you can follow a similar strategy to create amazing posts to promote the blog on social platforms. Finally, AI algorithms can begin to analyze user engagement metrics and conversion rates to help editors refine their content strategy. These insights enable data-driven decisions, leading to content that resonates with your target audience.

So, should AI be blindly adopted for content generation?  Wait, not so fast. While artificial intelligence offers numerous productivity benefits, it also comes with concerns about the reliability of the content produced. As Tom pointed out in the talk, “AI models predict the most likely next thing” but “they don’t know what they don’t know.” Because they can only have knowledge based on the data they have been trained on, their ability to predict the next thing is limited. This means that what they predict might turn out to be a lie.

AI response error cms

Consider the example above. Apostrophe’s CTO asked for 100 words on ducks in the style of Kurt Vonnegut with some links. The problem is that the American writer never covered that topic. Since the prompt included a request for a reference, the generative AI must have felt pressure to produce a citation. Thus, it generated a link to a PDF containing the “Duck Ford” story. Not only did Vonnegut never write this story, but the link pointed to a nonexistent site.

So, AI-generated content may contain bias, inaccuracies, and even factual errors. After all, artificial intelligence models are trained primarily with user-generated data, and we all know that our behavior on the Internet does not always represent the best version of ourselves. AI language systems may also have some trouble maintaining creativity. This is a direct consequence of the limited set of data they are trained on. In short, content produced by artificial intelligence should always be reviewed by a human.

Let’s now explore the effects of AI on a real-world CMS.

How AI Has Enhanced Apostrophe

Apostrophe officially supports AI through the ai-helper module. This enhances Apostrophe with AI-driven helpers, offering:

  • A menu option to generate rich text from a prompt via OpenAI.
  • A button to generate an image from a text prompt via OpenAI.

In other words, ai-helper offers a simplified way to integrate OpenAI functionality into an Apostrophe project with just a few lines of code. The only prerequisite to getting it to work is an OpenAI API key. This means that it is subject to OpenAI API pricing, which may change over time. 

Keep in mind that this module is still in beta, as AI systems can generate surprising and sometimes inappropriate results. Stay tuned for our step-by-step article on how to add ai-helper to an existing Apostrophe app.

See the rich-text generation feature in action in the clip below from the talk:

rich text clip

As you can see, you can write a ChatGPT prompt and get the result without leaving the CMS. The content generation process only takes a few clicks. Note that the resulting article includes headings and formatting. That is because the ai-helper module automatically converts the plain text produced by ChatGPT into Markdown. Cool, isn't it?

The other AI-based feature is the image generator button. See in action in the clip below:

Image generation

Again, after writing a short question, you get some images of DALL-E. To ensure that the AI-generated images are of high quality, you must manually review them before adding them to the gallery. If you would like to have some images similar to the selected one from OpenAI, simply click on the "Variations" button. Behind the scenes, Apostrophe will ask DALL-E to produce some similar images for you. Once you import an AI-generated image to the gallery, you can add it to the page like any other media file.

Here is what the final result looks like:

AI CMS response

Et voilà! In a matter of seconds and with just a bunch of clicks, you can go from a blank page to a complete blog post. This is the power of an AI CMS like Apostrophe.

How Apostrophe Plans to Fully Embrace AI

AI has been present in some status for more than fifty years, but the recent surge in advancements started only a few months ago. Things are evolving at a crazy pace, and what was once merely a dream has now become part of our everyday life. Getting on the hype train without a sound plan does not seem to be a good idea. Here is why Apostrophe has devised a solid plan to embrace AI the right way.

If you take a look at the Apostrophe product roadmap, you will notice that some feature requests and ideas involve artificial intelligence. These include using generative models to provide recommended SEO and accessibility metadata for documents in the CMS. This is just an example, but keep in mind that Apostrophe has an AI-focused roadmap ahead.

The Apostrophe team is always listening to the community. So if you have ideas about features you would like to have, let us hear from you on GitHub or Discord!

Conclusion

In this article, we looked at some basic definitions of AI and how it can change the world of content management forever. To learn how AI can support a modern CMS, we saw some features that the OpenAI APIs can add to ApostropheCMS, as told by the company's CTO Tom Boutell at the Philly Tech Week event.

With the innovative ai-helper module, you can have ChatGPT generate the text of an article for you. In addition, you can ask DALL-E to generate some creative images to feature in your piece. As proven here, AI can be very helpful in the right circumstances. Apostrophe has a structured roadmap to integrate the best AI features into your editing experience!