By Sabrina Farmer, CTO at GitLab
AI is becoming an increasingly critical component in software development. However, as is the case when implementing any new tool, there are potential growing pains that may make the transition to AI-powered software development more challenging.
AI has the potential to be a hugely transformative tool for software development, with benefits such as faster iteration cycles, fewer vulnerabilities, and less time spent on administrative tasks – all allowing organizations to ship software at the speed of the market. To achieve these productivity gains, organizations must consider making process- and culture-specific changes alongside adding AI-powered tools.
AI Training Gap
A recent GitLab research study found that 25% of individual contributors said their organizations do not provide adequate training and resources for using AI. In comparison, only 15% of C-level executives felt the same – highlighting a gap between how executives and their teams perceive investments in AI training.
This may be because some organizations still view AI as a potential replacement for developers rather than a tool that enables more creative and strategic human-centered work. To fulfill AI’s promise, organizations must supplement their investments in AI with investments in training and development resources that allow teams to build momentum and motivation over time.
Organizations should give their teams a grace period to determine how AI best fits their processes. Initially, productivity may decline as teams adjust to new workflows, but by testing how AI can best fit into their day-to-day workflows, teams will build trust in their new tools and see better results.
Toolchain Sprawl
One overlooked factor that can detract from developer experience and impact overall productivity is toolchain sprawl, or having multiple point solutions across software development workflows. GitLab’s research found that two-thirds of DevSecOps professionals want to consolidate their toolchain, with many citing negative impacts on developer experience caused by context-switching between tools.
Toolchain sprawl has additional drawbacks, such as adding cost and complexity, generating silos, and making it more challenging to standardize processes across teams. It also creates security concerns due to expanding surface attack areas and unnecessary handoff points. AI-powered point solutions compound these issues.
Leaders can adopt consistent, strategic best practices during this pivotal time of AI adoption rather than attempting to integrate AI into unwieldy and complex toolchains. Before incorporating new AI tools, organizations should evaluate their existing toolchains to determine areas where they can streamline or eliminate disparate tools to avoid the strain of integrating excess tools with AI solutions.
Outdated Productivity Metrics
Developer productivity is a top concern for the C-suite. Nearly all (99%) leaders surveyed said that measuring developer productivity could help business growth, but 45% aren’t measuring productivity against business outcomes. While measuring developer productivity has always been difficult, AI has compounded the challenge.
Many organizations struggle to quantify the impact of AI-powered tools on their developer productivity or accurately gauge how their results reflect real-world business outcomes. Traditional metrics, such as lines of code, code commits, or task completion, are insufficient when assessing development’s impact on a business’s bottom line.
The best approach to modernizing measurement practices begins with consolidating quantitative data from throughout the software development lifecycle with insights from developers on how AI is supporting or hindering their daily work.
To determine AI’s efficacy in software development, organizations should evaluate ROI based on user adoption, time to market, revenue, and customer satisfaction metrics. The most relevant business outcomes to monitor will likely differ across companies, departments, and projects.
AI has the potential to accelerate and evolve DevSecOps practices. Organizations can sidestep potential roadblocks and see faster productivity gains by proactively addressing the cultural and process-oriented challenges that may arise during the initial stages of AI implementation.
##
ABOUT THE AUTHOR
Sabrina Farmer is the Chief Technology Officer at GitLab, where she leads software engineering, operations, and customer support teams to execute the company’s technical vision and strategy and oversee the development and delivery of GitLab’s products and services.
Prior to GitLab, Sabrina spent nearly two decades at Google, where she most recently served as vice president of engineering, core infrastructure. During her tenure with Google, she was directly responsible for the reliability, performance, and efficiency of all of Google’s billion-user products and infrastructure.
A long-time advocate for women in technology, Farmer earned a B.S. in Computer Science at the University of New Orleans, where she established two scholarships to help level the playing field for inclusion and empowerment in technology.
