Generative AI in application development

Interesting facts

Over the past twenty years, I have mostly helped business areas that were not only users of enterprise software, but also contributed with their knowledge and will to a more efficient IT environment.

Custom software development is mostly based on the paradigm of the user describing to the developer as precisely as possible what he or she needs, but this cannot be achieved.

As the design itself is time-consuming, not to mention the implementation, delivery and deployment, users often do not wait for these to be completed and over the years build a solution themselves that effectively supports the day-to-day work of the company.
Of course, ignoring classic development principles can cause headaches in the long run, but there are now a number of low-code tools that allow you to build a solution that can be built on in the long run, rather than a huge excel spreadsheet in the absence of knowledge.
We use the Oracle Application Express has already proven itself as a user-friendly yet versatile framework that manages huge amounts of data and is popular with business users. As with little development knowledge, you can achieve fast and spectacular results that are accessible to the whole organisation.
In the near future, not only can individual regions, page elements and their properties be created without coding, but also SQL statements that collect the information itself.

Let's take a simple example:
We would like to create a chart in our application that displays the pseudo payments by department.
Launch the wizard to create the diagram.


Here we should give an SQL statement, but instead we will make the following request:
Show the average salary of employees by department".

The developer (who in our case may also be the user) can then generate the appropriate SQL statement by clicking on the "APEX Assistant" button.
APEX sends the text to a Generative AI REST service which returns the result as an Oracle SQL statement as shown in the example below.
Instead of having to start searching for the right tables, then linking them and performing the right operations, APEX Assistant does it for you.

Of course, it can also work by letting the user type in what they want to see, rather than clicking on buttons to modify the results of an existing report, as is already the case in a traditional BI tool.


The process is therefore as follows:
The developer - who in this case, because of the low-code approach, could be the user - creates a request fromt (prompt).
The request is processed by the built-in RAG (Retrieval Augmented Generation) module complements additional information - in our case from the Oracle database, which can be:- The names of the tables used in the APEX application- The attributes of the tables together with their data type- The primary keys of the tables, their identifiers- The foreign keys of the tables
The completed request (prompt) will be sent to the LLM (Large Language Model) processor
A Reply - in the form of an SQL statement - returned to APEX Assistant

 

Since all elements, pages, user habits, settings - as well as the data itself - are stored in the database and can be queried, there is nothing to stop the language model from providing even better answers to your questions over time.
All this points to a very exciting future not only for application development but also for the use of applications.


Using the APEX low coding requirements framework, the boundaries between developers and users, suppliers and customers, IT and business are increasingly blurred, creating a close collaboration that results in the most efficient IT solutions for all companies.
This article is based on a blog post by Oracle.
 

 

Sharing:

Blog post categories:

Discover how much more efficient it is to work with a single integrated system!

ADVICE

30-minute free consultation

Book an appointment for a free 30-minute consultation, where we will give you a deeper insight into how we can create a customized integrated system for you in less than 10 weeks!

en_USEN