Combining Developer Knowledge With Artificial Intelligence to Improve Software Maintenance

In our increasingly digitized world, enterprise software applications can better serve your business and your customers. By connecting every department within a company, enterprise systems allow companies to improve productivity and efficiency. These systems are generally created with specific goals in mind and serve many users at the same time, typically over a computer network instead of an end-user application.

From payment processing and online shopping to automated billing, interactive product catalogs, business process management, content management, security, and more, the services such software systems provide are built to satisfy the needs of businesses, schools, clubs, charities, and government organizations alike. These systems are frequently enhanced to meet the changing needs and opportunities of the particular entity for which they were written.

Given their critical importance, scale, and ability to support the mission of entire enterprises, enterprise software systems are complex and require specialized abilities and unique knowledge to update, add new features, fix bugs, meet regulatory requirements, address information security needs, and generally solve problems. When an issue arises, it should be addressed quickly to ensure it doesn’t spiral out of control and become a more serious issue that results in missed business opportunities. Consequently, ongoing enterprise-level systems maintenance is required as organizations grow and evolve, which costs your organization a significant amount of money. Generally, about three-quarters of an enterprise, ‘s IT budget for software is maintenance costs.

Unfortunately, between the global developer shortage and the brief average tenure of a developer at one job, organizations across industries are struggling to maintain their enterprise-level software systems in an effective manner. Complicating the issue, today’s code search tools, linters, and static and dynamic analysis tools tend to be inadequate when it comes to software maintenance.

Fortunately, employing a novel approach to artificial intelligence within software development tools can help save a significant amount of time and money while minimizing the risk associated with making changes in complex enterprise systems.

The Importance of Software Maintenance

While the development of enterprise-level systems is important, ongoing maintenance is just as — if not more — significant to the success of your organization. Regular software maintenance helps ensure trouble-free use and high performance, fewer issues, and rapid adaptation to changes and opportunities in the business environment.

Technical innovations are introduced constantly to improve the efficiency of the software itself and streamline business operations, but without regular software maintenance, enterprise-level systems will cease to use the latest technology and can simply become ineffective. Worse yet, the more complex the software, the more maintenance it will likely need to have to ensure optimal performance.

That means software maintenance isn’t just about fixing bugs — it also involves keeping things running the way your users expect. Consider how some of the four types of software maintenance go beyond simply fixing defects in your code:

1. Corrective maintenance: Performed to identify, rectify and correct any discovered problems, thereby restoring functionality to its intended form.
2. Adaptive maintenance: Performed when the environment of your software changes to ensure continued usability.
3. Perfective maintenance: Performed to add features that enhance the user experience and remove functionality that is not effective.
4. Corrective maintenance: Performed to detect and correct potential faults in the software before they take effect.

Whether correcting bugs, making the system compatible with a changing environment, removing obsolete functionality, or improving system performance, software maintenance is critical for optimal business growth.

A Skills Shortage or a Death of Knowledge?

Unfortunately, software maintenance is becoming increasingly difficult because it requires developers familiar with the software and the business operations. To become adequately productive and proficient, developers new to systems often require months — even years — of on-the-job training to avoid the mistake of making dangerous changes that put software at risk.

Complicating the issue, the Bureau of Labor Statistics indicates in the U.S. alone the shortage of software developers will exceed 1.2 million by 2026 and that more than a half-million developers will have left the market during that timeframe. Plus, about two-thirds of the respondents to the Harvey Nash/KPMG CIO Survey noted the growing skills shortage prevented them from keeping up with the pace of innovation.

The primary problem, though, is the lack of knowledge, not just the lack of developers. With the average tenure of a developer in a job being only a couple of years, organizations can no longer rely on developers to understand what the software does — under the covers. The specialized domain knowledge of the industry and enterprise that enables programmers to efficiently and effectively maintain and support complex enterprise software systems leaves when these developers depart.

The challenge is that discerning the true intent of functionality in code is not straightforward — and many modern tools are insufficient when it comes to helping developers find the specific lines of code that require attention.

Inadequacies of Today’s Tools

Today, developers spend roughly 75% of their time searching through source code to identify the code representing the functionality that needs to be updated. And if a software developer fails to understand how changing code in one area of the code impacts the system as a whole, even a minor tweak can break the entire system.

Since developers need to find the relevant code and build a mental simulation of how it works to safely make changes, code search tools, linters, and static and dynamic analysis tools are inadequate. Yes, these tools help developers improve their efficiency and effectiveness, but challenges remain since the simulation is left up to the developer.

Whether localizing bugs, repairing programs, or synthesizing code, modern tools can analyze million-line codebases, indicate possible errors, and even suggest where to look. But most tools that aim to help developers understand code automate only fragments of the overall cognitive process by merely identifying disconnected facts, leaving the developer to disambiguate the pieces and stitch together only the relevant ones to build a mental model of the behaviors within the system. As such, developers are forced to undergo the time-consuming, mentally challenging, and error-prone endeavor of piecing together that mental model to effectively modify complex and critical systems.

How Cognitive Automation Helps

Today, enterprises are beginning to leverage artificial intelligence-based tools that take a novel approach to AI to automate the identification of the specific lines of code that require attention — no matter how entwined throughout the system that code might be. Using AI in a novel way empowers the code repository to become a knowledge repository that represents source code in the same way that a human thinks about the world — in cause and effect.

The result enables cognitive automation, and these innovative AI tools do the actual thinking for developers, but are precise and generate significantly greater productivity gains than other AI-driven tools, meaning they deliver results that matter. Equipped with the code necessary to safely conduct software maintenance on an ongoing basis, human developers can confidently make changes in a way that protects the system from risk.

Understanding the vital importance of enterprise software maintenance, organizations across industries are attempting to overcome the programming language skills shortage by utilizing modern code search tools, linters, and static and dynamic analysis tools to improve developer productivity. However, forward-thinking enterprises are recognizing two things: First, software maintenance challenges are not the result of a developer shortage but instead a product of a lack of knowledge about the innards of the system. Second, many of today’s tools are woefully inadequate when it comes to achieving the developer’s goal. By employing AI-driven cognitive automation tools, enterprises are able to save time, and money and mitigate the risk involved in making changes within complex software systems.

Steve Brothers

Steve joined Phase Change as the COO in 2018, bringing over 30 years of experience in technology-related organizations with leadership, technical and sales roles in industries such as financial services, healthcare and services. Previously, Steve held positions as CEO at Ajubeo and Executive Vice President and CIO for Urban Lending Solutions. Steve graduated from the University of Colorado at Boulder and holds a B.A. in Philosophy and a B.S. in Information Systems. Steve is a proud father of two boys, is a mentor at Galvanize and resides in Golden, CO.

Follow us :