Thousands of coding languages have been used over time to create the applications that run our businesses, our government, our economy, and internet lives. While most are no longer used to build applications, enterprises continue to rely on decades-old applications that run on them, creating an unsolvable and fragmented problem for IT leaders. Legacy systems cost us almost half our time each week (or more) and billions of dollars. Yet in the rush to overhaul legacy systems, we are turning to illusory and costly quick fixes instead of breaking the mold for how we build software.
Business and IT leaders are looking to low-code/no-code and AI-assisted code generators as a quick fix for overhauling legacy systems or developing new applications. Those tools promise cheaper and speedier outcomes, tempting business leaders to believe that they can reduce their reliance on professional developers. Buyers beware: these quick fixes may result in a slow death spiral for your IT resources, creating even more uniquely coded legacy applications that the organization will need to maintain manually for the foreseeable future.
We need a completely new way to build software. That is a hard paradigm shift because code has been the foundation for our computing world. Those organizations that reject the cheap band-aid approaches and pivot to future-proof software architectures, which are free of legacy code and runtimes of the past, will be the true winners.
Consider your options carefully to avoid the next legacy code minefield. Low-code/no-code and generative AI solutions will create a fragmented landscape because they don’t take the application’s full lifecycle into account. It’s an easy choice to focus on application creation versus future maintainability, but it’s a decision that will impose significant future costs and complexities on your organization and potentially force yet another language onto IT personnel.
Ending our reliance on code in favor of maintainable applications, where the creation phase is collaborative between business and tech staff, is essential – even if that is not the solution being promoted by low-code/no-code companies. Until this is done, business and IT leaders will continue to focus on the proverbial tip of the iceberg, missing the much larger dangers lurking under the surface – expensive and almost impossible-to-maintain applications.
Chief Product Officer at Unqork.
The lure of quick and cheap is real
Budgets are getting smaller; inflation is driving prices up. Modernizing a legacy enterprise IT system is undeniably expensive. The U.S. Government Accountability Office’s review of various agencies’ IT systems shows that these are aged between 8 and 51 years and incur operation and maintenance costs of approximately $337 million annually.
Government or private sector, mainframes, data centers and legacy applications create massive amounts of technical debt – particularly for large enterprises that have been around for decades. These systems are difficult and expensive to maintain, causing engineers to patch applications that, at times, are older than they are. Looking for quick solutions like low-code/no-code may seem like a smart move at the time, if you forgo the longer-term implications. And most IT leaders are doing just that. Compared to 2022, Gartner forecasts low-code technologies to grow by 20 percent this year.
Unsustainable applications and technical debt
Low-code/no-code and generative AI can’t address the complex needs of enterprise applications, sustainably. Generative AI, when pointed at the right target, can do wonders. But technical debt and quick fixes to applications are problems stemming from code. Using those tools to create more code makes it worse.
Technical debt doesn’t only eat up budget, it takes time to contend with aging mainframes, data centers and legacy applications. Estimates for time alone to manage this debt suggests that developers spend ~42% of their week dealing with technical debt and bad code. While expense varies, the estimated impact of managing technical debt on global GDP is approximately $3 trillion.
Adopting low-code/no-code and generative AI solutions adds to the challenge of maintaining extinct languages and outdated technology stacks. The issue is twofold, generative AI has not been trained well enough to generate “good code”, and also creates uniquely coded, unsustainable applications that can be considered legacy from day one. When applied in this specific circumstance, the result is complex code that people can’t easily understand or use. It creates more problems than it solves.
Breaking the mold of how we build software
This is an inflection point. Consideration of the future of work and technology stacks looms large. It can only happen if we move towards data-defined software development, such as codeless architecture. With this approach, applications are described in open, flexible, and event-based definitions – not in unique code – which are then interpreted by an optimized run-time engine. By relying on reusable components, we eliminate the need to introduce new, uniquely-coded applications that add to our technical debt.
This paradigm shift will allow developers – not unlike moving to the cloud – to refocus more of their time on innovating rather than keeping the lights on, which in turn will drive time-to-market and productivity while accelerating technology backlog and capabilities. For IT leaders, data-centric application development will enable them to bring down total cost of ownership (TOC), while focusing on agility, efficiency, and scalability; not maintenance, patching and sunsetting.
