The current wave of organizations rushing to adopt Artificial Intelligence (AI) with a singular focus on speed is reminiscent of past technological transformations that prioritized velocity over fundamental value, leading to a dilution of their original intent. This myopic pursuit of speed, often a superficial interpretation of methodologies like Agile and DevOps, risks repeating historical mistakes and failing to unlock the true potential of AI. The core objective of software development and technological advancement should not be speed for its own sake, but rather the ability to gather early and frequent feedback to drive genuine user value and strategic agility.
The Agile Mirage: Speed as a Hollow Slogan
The Agile movement, intended to foster adaptability and responsiveness, has in many instances devolved into a mere pursuit of "speed." This shift has largely stripped Agile of its deeper purpose: to enable teams to iterate quickly, gather user feedback, and pivot their direction based on that feedback. When the primary goal becomes simply increasing the rate of change, organizations risk building more features faster, but not necessarily better or more valuable features. This can lead to feature bloat and constant flux, ultimately alienating users.
A compelling case in point is the evolution of word processing software. Microsoft Word, once the undisputed market leader, is now significantly outpaced by Google Docs. While Word boasts an extensive feature set, Google Docs, with its streamlined functionality and collaborative capabilities, has captured a larger market share. Data from 6sense indicates that in the third quarter of 2023, Google Docs held approximately 9.6% of the market, while Microsoft Word accounted for only 3.9%. This shift suggests that the perceived value for users often lies not in the sheer number of features or the speed of development, but in the compelling nature of selected features and the ease of use and collaboration. Organizations that focus solely on "straight-line speed" may be overlooking the fundamental principle of creating software that users truly value.
The AI Acceleration Trap: A Pattern of Disappointment
The current enthusiasm for adopting AI for speed echoes similar narratives surrounding previous technological trends. Over the past decade or two, many organizations have undergone "Agile Transformations," adopted "DevOps" practices, and launched "platform initiatives," all with the stated goal of increasing speed. However, a consistent pattern emerges: these initiatives often fail to deliver significant, sustainable improvements, leading to a sense of disillusionment and a search for the next technological silver bullet.
When leaders consistently announce initiatives for "straight-line speed" without achieving meaningful outcomes, it raises questions about their genuine intentions. While the allure of achieving "DORA Elite Performance" metrics or appearing technologically advanced might be a factor, the absence of a clear, fundamental reason for increased velocity—namely, the desire for early and frequent feedback—suggests a deeper issue. The repeated cycle of implementing methodologies for speed without achieving its core benefits points to a lack of understanding or commitment to the underlying principles of effective software development. Leaders who continue to champion AI as the ultimate solution for speed after numerous failed attempts with previous methodologies risk being perceived as deluded, failing to address the root causes of their delivery challenges.
The Feedback Metronome: Setting the Rhythm for True Agility
The true power of iterative development and technological advancement lies in prioritizing feedback over raw speed. When feedback becomes the central organizing principle, it dictates the pace of the entire software delivery process. This "feedback metronome" allows organizations to allocate sufficient time and resources to process this feedback, enabling the rapid course corrections that are the hallmark of Agile methodologies.
Organizations and teams that effectively leverage feedback as their rhythm setter are adept at identifying and eliminating work that hinders the feedback loop. They design their teams to minimize dependencies, streamline change approval processes, and empower teams to make deployment decisions and observe the outcomes. This approach is deeply rooted in established frameworks like the DORA model, which emphasizes a generative culture, transformational leadership, lean product management, and continuous delivery. These elements are not accidental; they are the culmination of decades of research and practice aimed at enabling teams to deliver value frequently and learn from the process. The speed achieved by these teams is a byproduct of their focus on frequent, high-quality feedback, which allows them to discover what truly needs to be built.
Team Elite: A Case Study in Feedback-Driven Delivery
The experience of "Team Elite" within a large healthcare company provides a compelling example of how prioritizing feedback can transform software delivery. This team was responsible for critical patient management and emergency triage software, where software defects could have life-or-death consequences. Previously, their patient management system was released every six months, with a two-week testing cycle for their decision support system, which could extend to another two weeks if issues were found.
Through a six-month program, Team Elite managed to shift their deployment frequency to a deployable software version every three hours. This was achieved not just by adopting strong technical practices, but crucially, by removing inefficiencies. The implementation of "specification by example" with executable specifications, coupled with the elimination of bureaucratic checking stages, proved more effective than simply adding new processes.
The tangible outcome of this transformation was significant. A critical deal with a new healthcare provider required the integration of a decision management API into their website. Team Elite successfully delivered a working API in just two weeks, leading to a live integration and securing a contract valued at $1.8 million (equivalent to $2.5 million in today’s currency). This success underscores the point that without a critical review and transformation of the route to production, simply introducing AI, or any other technology, will yield no substantial change, mirroring the limited impact of previous Agile, DevOps, or Platform Engineering adoptions. The fundamental step is to map the flow of value and address the broken elements, a process extensively documented in works like "Continuous Delivery: Reliable Software Releases Through Build, Test, and Deployment Automation" by Dave Farley and Jez Humble.
Rethinking AI Adoption: Beyond Mere Speed
The question "Why are you adopting AI?" is becoming increasingly critical. If the answer remains "to speed things up," it signals a missed opportunity and a continuation of outdated paradigms. A more mature and impactful approach to AI adoption requires a shift in focus towards frequent feedback and decision agility.
Organizations that have already mastered the throughput/stability trade-off through practices like Continuous Delivery are less likely to be solely driven by the pursuit of speed. Instead, they will seek more valuable opportunities enabled by AI. Two such opportunities are the formation of smaller, more agile teams and the development of more ambitious software.
The traditional "two-pizza team" model, while a pragmatic attempt to balance communication and coordination, is often still too large. With the assistance of AI, the ideal team size could shrink further, potentially to "one-pizza teams" or even smaller. Small, highly autonomous teams working on loosely coupled components are ideally positioned to leverage AI-assisted software development effectively.
Furthermore, AI can empower teams to tackle more ambitious projects that were previously considered too risky or complex. This could include undertaking significant globalization efforts or exploring innovative feature ideas that were historically hindered by a lack of clarity. AI-powered prototyping can provide the necessary exploration and validation that was previously unfeasible.
Ultimately, the most impactful use of AI will stem from a foundational improvement in software delivery processes and deployment pipelines, with a strong emphasis on closing the feedback loop. When this feedback loop is actively used as a metronome to set the rhythm for development, organizations will naturally seek more imaginative and value-driven applications of AI than simply accelerating existing workflows.
