GenAI and the Future of Software

Picture of John Doe

John Doe

Click edit button to change this text. Lorem ipsum dolor sit amet consectetur adipiscing elit dolor

Generative AI is disrupting the world around us. From writing articles, generating visual content, to creating software, it is taking over tasks with astounding speed. In the world of software GenAI takes a special place: it is the industry that gave birth to GenAI; and GenAI will change that same industry at a pace never seen before. But what does ik actually mean for the software industry? And for software developers? And can it build complex enterprise use case? This article tries to shed some light on these questions.

The software industry is used to disruption and evolution: it is the basis of its emergence and change is at the epicenter of its existence. Thirty years ago, I started my career on an IBM mainframe, a computer so large you could comfortably live in it with your family, while being less powerful than the smartwatch on your wrist today.

If you worked as a software developer on a mainframe, like I did, you would program your own IMS database, would need to allocate memory space for your variables (and free them, never forget!) and compiling your program took a night: in the morning you would find a printout of the errors on your desk. You would need to write Job Control Language (JCL) to control the sequence of program execution and you would live in the company basement, never to meet the users of your programs in real life.

That was 30 years ago.

In the meantime, mainframes got out of fashion, client-server technology became the norm, and now we are moving to the cloud, with virtualization and containerization. From a business application perspective, we moved from mainframe to ERP to SaaS to GenAI; from a software perspective we moved from 3GL (Cobol) to 4GL (Java) to low-code (OutSystems) to GenAI.

Software developers, don’t worry!

This evolution is a great thing for developers: programming IMS databases is horrendously difficult, stupendously time-consuming and exceptionally boring, so when relational databases came into fashion, I was relieved; when Java and .Net languages replaced procedural programming and introduced object- and data-oriented programming, I was a happy person; and when low-code got rid of tons of boring, repetitive coding, like variable binding, I was overjoyed. With every improvement down this evolutionary path, I could spend more time analyzing instead of developing, and it slowly got me out of the basement into the open air, and more in touch with the business users of the products I build.

Now there is a new evolution: low-code and GenAI, the most powerful software combination the world has seen so far. GenAI will not just create some screens and some logic. It will be able to generate complex applications, including screens, databases, logic and integrations, in minutes, by reading and analyzing requirements you feed it, for instance a Word document. What’s more, GenAI doesn’t stop at initial generation: it will be able to understand the complexities of the apps it created and change these apps when required, based on prompting and design documents.

Estimates are that this will speed up application creation in such a way that development cost will become a fraction of the TCO of software. And while some developers are scared AI will take their jobs, the opposite is likely to be true: because software creation will be so easy and cost-effective, there will be an explosion of the amount of software, from single-use software to custom-generated ERP systems. Because there will remain areas that AI does not understand, like proprietary back-end integration with an ERP landscape, there will still be high-quality work for software developers. And because of the sheer volume of code that is expected to be generated by AI, there will be a need for more software specialists.

Low-code takes center stage

An important question to answer is: why should low-code prevail and not high-code, like Java or C#? The obvious answer is that, with the use of Copilot or other AI high-coders, AI will create highly complex, hard to read and to maintain high-code, that would require an army of experienced developers to control. Simple programs are fine, but imagine you create a mobile banking application using Copilot, that would most likely result in tens of millions of lines of code. That code base needs to be able to manage the workload of millions of users executing complex tasks every day. If anything goes wrong, or there is a performance bottleneck, you will need some serious seniority to find and solve the bugs. And because high-code AI will replace most of the coding, experienced high-code developers will become even scarcer than they are today.

Low-code does not require that level of seniority: the model generated is much easier to understand than high code, so staff reaches the level of experience to support generated apps much quicker and less people are required for support. Because of the expected exponential increase in software generation, it is imperative that new, capable staff can be trained and onboarded quickly. Low-code fits this perfectly and is therefore the only viable option to support this.

OutSystems + AI has no limits

There is only one low-code platform out there now that is this far advanced using GenAI: OutSystems. For years they have been working on this and this year they released Mentor, their so-called AI digital worker, for early access.

GenAI and OutSystems opens up a whole new world of possibilities in the enterprise IT landscape:

  • Imagine you can use AI to create your current Salesforce or ServiceNow instance? Not just with the same functionality, but tailored to your business needs, so it fits like a glove? The business case is compelling, if you can save on these premium license and hosting costs dramatically.
  • What if you could reduce your ERP core to the bare essence, and generate the other modules using GenAI + OutSystems? That would give you a new set of option. For instance, you can move to S/4HANA with a much smaller core, which will be a straightforward and cost-effective move.
  • Or even take it a step further: if the core is so small and back to standard, you could keep running your current SAP Netweaver stack and NOT move to S/4HANA. This will save you a huge effort in terms of time, resources and money. If you are worried that your system will not be supported in the future, there are companies that offer maintenance services for SAP Netweaver beyond 2030.

Get ready for the future

Big, exciting things are afoot in the software development world. A new era is upon us that will revolutionize the way code is created. Developers shouldn’t worry about their jobs, instead they should adapt to this new reality and focus on how GenAI is going to change their work. And enterprises should get ready for the new, limitless possibilities this will generate. Personally, I see a lot of opportunity for both developers and business users, and I can’t wait to see this technology take us further into the future.

 

 

Original article can be found here.

Share this post