Machine Learning per Mouse Clickhttps://appnavi.eu/wp-content/uploads/2022/08/AppNavi_News_Product_AI-Search_EN.jpg19201080Daniel BecksteinDaniel Becksteinhttps://appnavi.eu/wp-content/plugins/ultimate-member/assets/img/default_avatar.jpg
New programming concepts require rethinking
Modern software applications are outdoing each other with ever more extensive possibilities. Developers and programmers are digging deep into their bag of tricks and breaking up familiar structures in order not to lose sight of the big picture themselves and to be able to develop ever more complex tools more easily. This is even giving rise to new trends in application development, as is currently becoming apparent. Keyword: Shadow Root or Shadow DOM. Here, elements are encapsulated and hidden from the actual tool in order to save resources.
Much to the detriment of digital adoption platforms. But why? To understand this, we need to take a look behind the scenes, i.e. what actually happens technically in a digital adoption solution.
What happens under the hood with digital adoption
Digital adoption platforms, such as AppNavi, when used in route creation, search the programming code of an application for appropriate elements. Such an element might be a button to click or an input field where something needs to be entered. This is the core of guiding users through applications. In this element recognition we speak of “search”.
The relocation and even hiding of application modules by developers causes difficulties for common detection and search algorithms. That is why we have completely revised our algorithm and adapted it to the new trends.
The new AI Search
The new AI Search is a further development of our previous Fuzzy Search for the recognition of elements in corresponding software. The core is still the fuzzy score theory, which gave the fuzzy search its name, but the new search follows a completely new approach and offers additional new functionalities, such as machine learning.
In addition to a new search and element recognition logic, it is now possible to actively train the algorithm. This means that you don’t have to rely purely on the algorithm, how exactly it recognizes elements in the code, you can (if necessary) actively detect elements via a superimposed interface and increase the probability of identification (is even specified in percent) and this simply by clicking the mouse without programming knowledge. With AppNavi this is then called “Update capture”.
The implementation of the new search does not mean that the fuzzy search will be replaced. Quite the opposite! The AI Search is added and you have the option to choose between both options in the system. Simply put, choose the search that works best for the desired result.
Due to the fact that AI Search follows a different ideology compared to Fuzzy Search and thus has to sift through less data, the functionality has also become much faster. Win-win – once again.
This also supports Competence Centers in particular, i.e. individual groups of people who look after a large number of AppNavi-supported applications and create countless routes. With the AI search, the creation becomes even easier and primarily more sustainable, as the content thus reacts much more flexibly to changes.
How the new AI Search works
The functionality of the new AI Search fits seamlessly into the proven AppNavi interface.
> A route is recorded as usual using the AppNavi Extension and the corresponding hotkeys. > It is always recommended to run a short test with the route preview. If a deviation is then detected, the new learning method is used. > In the route editor, the corresponding element can be specified again with the AI search to increase the probability of identification. Alternatively, it is possible to manually choose (via slider) between text or position, depending on which is more likely to apply. This can be especially helpful when values in a system or names in a location are constantly changing, such as CRMs or databases like ServiceNow or Salesforce.
This ensures high stability of routes and low maintenance. This quickly elevates users to a level that does not require any expertise at all, and the content is integrated into the application in a more future-proof manner through the layer.