Via standardization and AI to personalized documentation

In today's world, terms such as user experience (UX) and customer journey, and thus users, are increasingly becoming the focus of discussion. The main goal of technical documentation should be to provide a clear and understandable presentation of complex content that efficiently supports the execution of tasks. Especially in complex scenarios, it is a challenge to design software so intuitively that it can do without documentation altogether.

As a software developer, I personally prefer quickly skimmable, i.e. scannable, information to complete my tasks in complicated applications. Nevertheless, there is certainly a large number of user who can and want to absorb information in a different, much more efficient way. This is not the only reason why a general consideration in the context of the UX concept is recommended.

As a company specializing in software solutions in the field of technical documentation, we are naturally doubly challenged to address the issue of personalized software documentation. Where possible, we try to integrate new approaches if they fit our basic understanding.

Our basic understanding is clearly based on the topic of automation, and it stands to reason that we would like to avoid manual work - and that includes editorial work - wherever possible, especially since we also have to deal with the resource of labor in an increasingly sustainable manner.

Additionally required skills

Es gibt viele Ansätze, die bei der Dokumentation von Inhalten über die Jahre verfolgt wurden. Wir hatten Konzepte wie Simplyfied Technical English, Information Mapping, Funktionsdesign, sind in Praxis bei der Topicorientierung angekommen und diskutieren neue Ansätze wie das Story-Telling.

All these concepts require special skills. For example, I see storytelling more as a discipline that is located in marketing. This is a skill that will now probably also be needed in technical editing if this concept is to be pursued accordingly. Not only, but especially when writing software documentation, an in-depth knowledge of the respective domain is essential in order to be able to write high-quality instructions. Furthermore, numerous additional competencies are required that today's technical writers must possess in order to bring software to the users an understandable way.

In view of the existing shortage of skilled workers, one must ask whether the implementation of new approaches that presuppose an expansion of the skills already available is actually purposeful. Especially in light of the fact that we can't really afford to keep producing new legacy content, and most of the time we don't have the resources available to rework and refurbish it.

I think it's more important to use the expertise in the editorial team, and especially their view of users, to make the application more intuitive, to constantly improve the UX, and to ensure that less documentation is needed overall to be able to use an application. Reducing complexity must always be a priority. This requires the integration of technical documentation into the development process, making it part of the value creation process and thus contributing to the company's bottom line. Who in the company knows users better than technical writers?

But how do we get our personalized content now? By trying to make our content creation and delivery processes digital, or at least thinking about them digitally. We have to relieve employees, save resources, act sustainably and holistically, and thus further standardize and automate.

Personalization

Especially in modern software, users are known to the system. (I had already talked about LinkedIn on the subject of gendering ). As a user of complex software, I would of course appreciate it if the documentation supported me as purposefully as possible in my tasks, if the software is not already able to do this intuitively due to the complexity involved.

We live in the year 2023, in the age of artificial intelligence. If it is already necessary that I need appropriate documentation for a software, then I would like to know this in such a way that the contents are prepared exactly according to my needs. Maybe a short summary is enough for me, or I need an explanation in another specific context. technical writers can no longer do this during the creation process, even if the appropriate budget and manpower were available.

We definitely don't have time to experiment with new documentation approaches, since we usually don't start from scratch and of course have a corresponding documentation inventory to show for it, which then requires revision. In the future, the focus must be on providing information and networked knowledge in as structured a manner as possible.

Artificial intelligence should be our tool to prepare this information exactly when and how it is needed. In the future, the content will be prepared on demand by an AI and compiled in the way users need it. ChatGPT already shows what this looks like in a very impressive way. It is possible to have texts summarized and presented in a different way, even as a story, if users so desire. And of course, in the future, our software will know how we would like to be supported.

Standardization

I appreciate the consistency in the style and structure of a documentation because I can quickly find my way around at any time. On the other hand, we need lean processes that free up resources both in terms of manpower and financially. At c-rex.net, when it comes to standardization, we focus especially on specialized DITA and coordinated automation processes.

However, the point about standardization is not really the format such as DITA and XML or managing the content using a CCMS. At its core, it's about finding the optimal path for your own business. Yes, there are indeed companies that use simple tools like Word to create excellent, user-friendly documentation and manage their content efficiently on the file system. This works when the processes are standardized and defined. With tools such as c-rex.net DTS, for example, we meet the core requirements for modern documentation at all times, for example the seamless transformation of such content into formats such as HTML5 or even iiRDS in the context of Industry 4.0.

The real goal is a paradigm shift in documentation, a change of mindset. We need to standardize and automate processes in such a way that they can be seamlessly integrated into a uniform workflow. Metadata and the linking of content and other entities play a crucial role here.

In sum, this is how we generate knowledge. Knowledge that we can make available to AI systems, which in turn make it individually accessible to a wide variety of interested parties.

Machines also want to be informed

And what we keep forgetting in the discussion is that we don't just write for people anymore. We also write for machines, which is a piece of software that consumes editorial content - possibly in a completely different format - and may even configure itself as a result.

For example, we configure our DTS pipelines using only the process documentation or pull API documentation directly from the content API. How we do that is already described on our blog or Youtube channel and more will follow.

What about the translation then?

If we've already gotten to that point and realized that we're going to have content delivered to us individually in the future, we obviously need to make sure that our content is available in the languages of our users. As a consequence of AI providing information, this ultimately means that we can also have our content translated by machine (as always: provided the terminology is fed into the process).

By the way, if we follow the story-telling approach, we can no longer go through the classic translation process via a TMS anyway, if we are really consistent. Why? Because storytelling wants to convey emotions and empathy, for example. This is not possible with a segmented tabular translation, since translators are by definition tied into a predefined tabular corset.

(Preliminary) Conclusion

In the past, we have taken different approaches to how we create documentation: from information mapping to functional design. Now we may even have landed on story telling, because we want and of course have to put the user and customer journey more at the center. Today, this is state-of-the-art.

But are we still on the cutting edge when we try to address individual needs of our users with static content? I don't think so. Our goal must be to provide comprehensive information and link it both to each other and to external entities, such as software components or functions. Such links and metadata are what allow us to efficiently guide users and provide them with exactly the information they need - preferably before they know they are about to need it.

If we really want to provide users with personalized content, we can't do it with static, editorially created texts. users should be able to choose for themselves how they consume the information (just as readers on our website can decide for themselves at any time whether or not they want to consume our content in a gender-neutral way).

Since personalization of content can no longer be realistically implemented editorially, our only option is to use artificial intelligence. That is why standards and standardized procedures are more important than ever. But it also means that we can continue to write technical manuals in an informative way and choose a concept that suits us best. Ultimately, it will soon no longer be in the hands of technical editors how users use documentation.