Oracle APEX Generative

Development
October 2, 2024 | 8 minute read

Michael Hichwa
SVP, Software Development
This is a syndicated post, view the original post
Low-Code and Generative AI
Generative AI's productivity promise is exciting, but not all generated artifacts are created equal. The
reams of code streaming from large language models are impressive, but enterprise apps get
enhanced over decades. A generated jump-start is only valuable if teammates can understand and
evolve it over time. Model-driven, low-code tools offer a higher-level of abstraction by capturing
application intent in metadata. Developers use visual tools to interactively assemble pages from pre-
built components, specify the data to work with, configure various display options, and impose
conditional and dynamic behavior. The platform directly executes these developer definitions to let
end-users get their jobs done from desktop and mobile devices. Low-code app definitions with a
simple text format are succinct, easy to understand and compare, and simple to evolve as new
requirements arise. Targeting higher-level abstractions like these with generative AI is a superior
option. That approach reduces technical debt by keeping generated definitions human readable and
evolvable.

Generative Development with Oracle APEX

Oracle APEX is core to generative development for enterprises (GenDev), introduced by Juan Loaiza
at CloudWorld 2024; an AI-centric data infrastructure first introduced with Oracle Database 23ai,
offering a low-code application platform that lets you build scalable, secure web and mobile apps with
world-class features to deploy anywhere. It is model-driven by design, with app definitions that have a
simple text format. Oracle APEX’s AI Assistant and developer-centric features let you use natural
language to help generate SQL and application blueprints, and let end-users ask questions in natural
language in a chat dialog. Combined with other key technologies in Oracle Database 23ai such as
JSON Relational Duality Views and AI Vector Search, APEX is part of GenDev, an innovative
development infrastructure to accelerate the benefits of AI while helping to mitigate its risks.

Using Natural Language to Help Generate SQL

Enterprise app developers use the declarative SQL language to express the intent of what data is
required to retrieve, create, update, or delete for a given use case. While SQL's concepts are simple, in
a large enterprise data model with many tables and relationships, developers could use some help
remembering all the table and column names and join conditions to produce the data required for the
task at hand. In Oracle APEX, wherever you can type SQL you can summon the APEX AI Assistant to
help you generate any required SQL instead by explaining what you need in natural language. While
creating a chart to visualize the top 10 classrooms in use at a university, in Figure 1 the developer
asks a question in the conversation pane on the right. “SM” requests the top 10 classrooms for the
current calendar year offering the most courses. APEX AI Assistant can help generate the
corresponding SQL that you see on the left. The query joins tables for buildings,
classrooms, and courses, with where, group by, order by, and fetch first 10 clauses.

Figure 1: Using natural language to help generate SQL with APEX AI Assistant
To fine-tune the result, just ask APEX AI Assistant to help adjust it. For example, in Figure 2 notice in
the conversation on the right that the developer asks to concatenate building and room in the query.
This helps produce the updated query on the left.

Figure 2: Asking APEX AI Assistant to help refine an existing SQL query

With this kind of help, providing end users the Top Classrooms in Use page with the chart shown in
Figure 3 is a simple matter of describing the data to visualize in natural language and choosing a chart
type.

Figure 3: Page with bar chart visualizing results of SQL query for Top Classrooms

Using Natural Language to Help Generate an

Application Blueprint
To quickly jump-start application creation, you can explain the app you need to build in natural
language. APEX AI Assistant can help generate an app blueprint as shown in Figure 4. Again, you can
iterate on the prompt until the proposed blueprint is perfect, and then click (Create Application) to
help get the job done. This generates a set of application and page definitions, with declaratively
specified data sources. These crystallize your intent into understandable, evolvable metadata. From
there, you can proceed to adjust the definitions to add functionality that only you would know how to
build. No generated code to understand or worry about breaking!
Figure 4: Creating an app blueprint using natural language with APEX AI Assistant

Adding an AI-Powered Chatbot to Your

Application
When your app needs an AI-powered chatbot, it's easy with Oracle APEX. Simply configure a button
to open the chatbot dialog using the built-in "Show AI Assistant" action, provide the system prompt
that gives the chatbot its mission, and enter a welcome message. In short order, your app's end-users
can be asking questions to a chatbot assistant like the Sequoia University Assistant you see
in Figure 5. Using simple SQL statements, it’s easy to include your own application data into the
chatbot's system prompt to let the AI service generate even more useful and specific answers. This
technique is known as retrieval-augmented generation (RAG).

Figure 5: End users can ask questions to an app-specific AI-powered chatbot

Benefits of Generative Development with High-

Value Components
A key strength of Oracle APEX is its industry-leading set of highly functional, data-savvy
components. These are well suited to generative development because you simply configure their
data, behavior, and display options using high-level metadata. While the full set of components also
includes forms, reports, grids, calendars, maps, cards, trees, and more, three of APEX's most popular
components include:
 an interactive tabular report
 an e-commerce-style faceted search experience, and
 a versatile chart to visualize data of any kind.

Interactive Report Built-in Component

The Courses page in Figure 6 features an Interactive Report component. You start by configuring the
data source using a table name or SQL query, possibly generated from natural language with the help
of the AI-powered APEX AI Assistant. And that's it. This one configuration step taps into a wealth of
features that end-users love. The Interactive Report lets users view data in a tabular format, choose the
columns they want to see, filter the data to find a subset they are looking to focus on, format the data
with break groups and conditional highlighting, sort, aggregate, or derived computed formulas based
on the data, chart and group the data, and pivot the rows into columns when useful. Users can save
one or more named reports to re-run later, download the formatted results in multiple formats, email
the report to a colleague, or subscribe to receive in email a periodic report run. Developers have never
worked so little to offer so much rich, functionality to their end-users.

Figure 6: End users customize tabular reports and save as needed to run again later

Faceted Search Built-in Component

The Course Search page in Figure 7 shows an example of an e-commerce-like faceted search
experience. It was created in a few clicks of the Create Page Wizard. It only required choosing
the COURSES table as the subject of the search and reviewing the intelligently-selected set of columns
to use as search facets. Using a process familiar to anyone who has done online shopping, end-users
narrow the results by clicking one or more options in any combination of search categories. Oracle
APEX handles filtering the facets to show only applicable, remaining choices, and provides counts for
each one. By configuring an action on the Cards region displaying the results, you could let the end-
user drill down to get more information on a particular course.
Figure 7: End users search using examples from categories like an e-commerce site

Flexible Chart Built-in Component

When your app needs to visualize data of any kind, Oracle APEX lets you do that quickly. The chart
component supports all the different kinds of chart types you see in Figure 8. Just configure the data
you want to see. As you saw above, you can do that using natural language to explain what info needs
summarizing over what time frame and along what dimensions and the APEX AI Assistant can help
create the SQL that the chart component uses to retrieve the data to display.
Figure 8: Thumbnail examples of available chart types
These are just three of the most frequently used data-savvy regions Oracle APEX developers have in
their toolbox. However, this familiar approach works with any component. Whether configuring an
interactive report, faceted search, grid, form, app search, map, chart, or other kind of region, you
configure your intent in the same way. Simply indicate what regions and items to add to your page,
configure what data they should work with, set the display options to get the result you want, and
configure any interactivity the page requires.

Oracle APEX: A Key Part of GenDev

Oracle APEX is a strategic platform for the generative development of enterprise apps. Whether you
and your colleagues are low-coders or pro-coders, you can all collaborate using Oracle APEX to
quickly describe, iterate, and deliver sophisticated, data-centric enterprise solutions with AI assistance
throughout. Using both natural language and declarative builders, you capture the intent of your
business solutions and the data they require, staying focused on the WHAT instead of the HOW.

More Resources and Getting Started

Learn more about GenDev and how you can benefit from this technology here.
Visit apex.oracle.com to get started today with a free evaluation workspace to try out every Oracle
APEX feature, and if you like what you see you can provision an Always Free Oracle Cloud tenancy
with Oracle APEX running on Oracle Autonomous Database as the next step. From this page you can
also explore hands on labs, training and certification opportunities, and see the network of Oracle
APEX partners and community components produced by the worldwide Oracle APEX developer
community that is over 800,000 strong. You'll be in great company!
Michael Hichwa