Transforming the SDLC: Real-World Examples of How WillowTree Enhances Software Development With AI

Forecasts on how generative AI will revolutionize the software design lifecycle (SDLC) are staggering. A recent Gartner report projects that by 2027, AI-powered coding tools will be utilized by 70% of professional developers. This data aligns with a KPMG survey in which 73% of 300 global executives identified AI and machine learning as key drivers for their organizational growth, with many already exploring innovative ways to incorporate genAI into their coding strategies.

Although AI’s perpetual hype cycle can feel panic-inducing, we recommend an orchestrated approach over one motivated by FOMO, especially if you’re just beginning to make strategic decisions. At WillowTree, our expertise isn’t theoretical; we’re actively leveraging artificial intelligence to transform how we build software.

In this first installment of our series on using AI to augment the SDLC, I’ll delve into three distinct areas where AI empowers human intelligence and significantly enhances the value we deliver to our clients:

1. Efficiently analyzing and gathering requirements,
2. Unlocking creativity while reducing errors, and
3. Reducing friction in documentation and onboarding.

I hope our practical, daily application of AI provides inspiring use cases as you consider next steps.

Empowering Human Intelligence

Since our origins in 2008, WillowTree has supplied our developers and delivery experts with powerful tools that eliminate repetitive tasks and accelerate creative problem-solving. Generative AI is one of the most dynamic toolsets to date, and we trust our expert team to manage the critical reasoning, strategic planning, and high-level thinking necessary to deliver software on time and on budget.

By thoughtfully integrating AI tools throughout our workflow, we've created a development ecosystem that encourages automation while significantly enhancing our team's capabilities and upholding WillowTree's uncompromising quality standards.

1. Turbo-charging efficiency in analysis and requirement gathering

Let’s kick things off by spotlighting a noteworthy enhancement in efficiency during the initial phase of the SDLC — our deployment of artificial intelligence has slashed initial analysis time by up to 40%.

We achieve this by using enterprise AI platforms like Fuel iX, which boost efficiency and speed.

• Fuel iX automates the analysis of extensive text documents by identifying critical phrases, terms, and patterns that make it easier to define functional and non-functional requirements.
• We also use it for sentiment analysis on feedback from users and stakeholders. This approach helps us gauge the importance of different requirements and prioritize features based on user value.

I spoke to one of our product leads, Jes Fitt, to learn more about the everyday impact of her genAI tooling. She shared a compelling use case for generative AI when drafting stories and acceptance criteria. Without AI, Jes estimated that drafting five new user stories might take 15 to 20 minutes. However, with AI assistance, she can get her work across the finish line in closer to 2 minutes.

GenAI reduces Jes’s repetitive ticket-writing time by up to 300%, so she spends less effort on initial documentation and more on valuable activities such as subject matter expert (SME) interviews and requirement validation. At scale, time savings like these have a ripple effect, improving subsequent collaborations by freeing teams to reason and refine generated requirements with clients.

2. Unlocking creativity in development and reducing errors in coding

Another clear win for us has been using AI for algorithm generation, particularly math-based formulae like those required for bit operation and color shifting. Our mobile and web engineering teams are especially fond of AI support in such cases since they rarely encounter these formulae in the UI and front-end realms. Through dogged, prompt engineering, our teams can streamline the research, testing, and validation processes built upon more formulaic algorithms.

Overall, our software development teams report that AI tooling makes them 25 to 40% faster in critical aspects of the SDLC, such as scaffolding, implementing boilerplate and repetitive code, and data validation. These time savings unlock more space for the expert reasoning required for a well-engineered codebase. And this is where an AI-enhanced approach to software development becomes truly exciting.

AI unlocks creative freedom, allowing developers to experiment and innovate the pioneering solutions our clients count on from us. The ability to swiftly prototype a feature or algorithm equips our teams with early indications of the "right path," expediting their progress toward a more comprehensive and robust solution.

One of our iOS engineers, Nathan Molby, built an AI copilot to convert older XCTest-based unit tests into newer Swift Tests. While Nathan trained this copilot on Swift Testing documentation, he pared it down to a syntax set uniquely applicable to WillowTree use cases. The copilot also authors new Swift Tests from scratch against any code provided. Leveraging this copilot has allowed our teams to take swiftness to heart — cutting testing files by half and creating parameterized tests with fewer cycles.

3. Reducing friction in documentation and onboarding

AI tooling has also significantly streamlined the labor-intensive software delivery process at WillowTree, leading to enhanced documentation quality, more efficient project onboarding, and a smoother workflow.

Curating and refining content can be an absolute time suck, so we use Fuel iX and other conversational tools to reduce friction in maintaining a wiki or updating a README. These tools adeptly manage varied formatting, enabling team members to input documents from various sources and quickly merge them into an updated resource. During a recent retrospective, I witnessed a team member use Fuel iX to transform an unformatted list of action items and feedback into a comprehensive Markdown-based output that they subsequently merged into the team’s README. It happened in a matter of seconds.

Overall, our teams are making significant improvements in onboarding:

• As an agency, we frequently transition team members between projects. Enhanced documentation at fingerprints and other critical SDLC milestones enables faster alignment within a team, allowing collaborators to focus on crafting effective solutions.
• We develop AI copilots equipped with specialized domain knowledge. These copilots lessen the time experienced team members require to onboard newcomers, often cutting that time in half. This expedited process also reduces the input needed from other team members, enabling everyone to get heads-down and into their more meaningful work.

Looking forward: New tools for new use cases

If I've successfully illustrated how AI energizes WillowTree’s distinctive approach to software development, you're already contemplating opportunities for problem-solving, friction reduction, and boosting creative confidence within your organization.

Wins like these can serve as a launchpad for SDLC excellence, and AI advancements in software engineering are evolving in real time. We’re also discovering opportunities for improvement in the testing, deployment, and maintenance aspects of the SDLC. I’ll discuss these topics and more in subsequent installments, so stay tuned!

Until then, if you’re interested in learning more about how our AI-enhanced development expertise can optimize your next project, reach out. We're continually identifying novel ways to utilize AI tools, substantially improving development speed, code generation quality, and team satisfaction. We look forward to discussing how we can help you achieve your goals more efficiently.