02.11.26 By Dominick Profico
Over the last few years, the software industry has been trying to do something very familiar: employ new technology and make it fit the way we already work. As generative AI began showing up in real products and real workflows, most teams kept the same delivery models relied on for years. Same Agile ceremonies, sprint cycles and assumptions about how work moves from idea to production.
At first, that felt reasonable. Agile has served us well for nearly two decades. But as we have spent more time engineering products with AI, and talking to others doing the same, it became clear that we may be forcing AI into a model that was never designed for it. This is not because Agile itself is no longer relevant, but because the ground underneath it has fundamentally shifted.
“Across organizations, we’re starting to see that building faster horses isn’t the answer when the race itself has changed. Agile helped us scale innovation for a long time. But intelligent automation, adaptive systems, and generative design change where the real bottlenecks and risks are.”
– Dominick Profico, CTO of Bridgenext
This post is the first in a series that starts with an uncomfortable possibility: that Agile, at least as we practice it today, may be reaching the limits of its usefulness in an AI-driven world, and that acknowledging this might be the most constructive step forward.
We have spent years optimizing our stand-ups, refining our backlogs, and perfecting our sprint velocities. Yet, despite these efforts, the industry faces a staggering reality. Research from the Consortium for Information & Software Quality (CISQ) indicates that poor software quality costs the U.S. economy upwards of $2.41 trillion annually, with roughly $260 billion of that attributed specifically to failed projects and operational failures.
You guessed it: many of these teams had been “doing Agile” for years. If frameworks alone cured waste and unlocked quality, that $260B curve would have shifted in a positive direction by now. In reality, an efficient process does not always equal effective outcomes, something we often chuckle about in retrospect with clients who, as they put it, “did everything by the book…and still ended up rewriting the book.”
Generative AI exposes this flaw mercilessly. When an AI agent can generate code, write tests, and populate documentation almost instantaneously, many traditional Agile rituals begin to appear less like disciplined practices and more like performative work, ineffectual activity during a systemic disruption.
Traditionally, the bottleneck in software delivery was execution. Writing clean code took time. Testing took time. deploying took time. Agile was designed to manage this scarcity of execution bandwidth by breaking work into small, manageable chunks.
AI flips this dynamic on its head. Execution is becoming abundant and nearly instantaneous. When software “writing” is commoditized, the constraint shifts upstream and downstream. The bottleneck is no longer “how long will it take to build this?” It is now “should we build this?” and “is what we built safe, ethical, and aligned with our strategy?
In a world of abundant execution, the value of a developer or product owner shifts from output production to strategic orchestration. We don’t need a two-week sprint to determine if a feature is viable if an AI can prototype it in an hour.
This renders many standard Agile ceremonies obsolete. Why hold a daily stand-up to discuss yesterday’s coding progress when the code was generated and tested overnight by an autonomous agent? Why spend hours on story point estimation for tasks that take minutes to execute? The rituals that once provided clarity now risk introducing latency.
Agile frameworks, in their current form, are ill-equipped to handle the risks introduced by AI. Traditional Agile emphasizes speed and adaptability, often at the expense of rigorous, upfront documentation or long-term architectural planning. “Working software over comprehensive documentation” is a core tenet.
However, when AI generates code at scale, “working software” can quickly become a liability if it isn’t governed correctly. The promise of hyper-automation brings severe risks:
Current Agile methodologies do not account for non-human actors in the workflow. They assume human agency, human error rates, and human communication speeds. They lack the native governance structures required to audit AI agents or verify the integrity of synthetic codebases.
Simply adding AI to our Agile teams toolbox is insufficient. We cannot just use AI to supercharge stand-ups or automate backlog grooming and call it “modernization.” That is the equivalent of putting a jet engine on a stagecoach.
We need a new operating model that prioritizes strategic orchestration and ethical governance over task management.
The argument that Agile is “dead” is not a dismissal of its principles, collaboration, adaptability, and customer focus, these elements remain vital. However, the mechanisms of Agile, the sprints, the points, the rigid ceremonies, are artifacts of a time when writing software was hard and slow. That era is ending.
We are entering a phase where the creation of technology is fluid and fast. To survive, we must stop clinging to the rituals of the past and start architecting a new way of working. We need a framework that recognizes AI as a fundamental shift in how software is built and delivered.
If we don’t disrupt our own processes, the market, and the $260 billion in waste we continue to generate, will do it for us.
This is part one of our series on Agile in the AI Era. In the next post, we will explore the counter-argument: how AI might be the ultimate enabler of the Agile promise.
References
www.it-cisq.org/cisq-files/pdf/CPSQ-2020-report.pdf
www.forrester.com/blogs/amidst-the-ai-hype-agile-still-remains-relevant-in-2025/
www.gartner.com/en/articles/devops
www.deloitte.com/us/en/insights/topics/digital-transformation/ai-tech-investment-roi.html
www.idc.com/wp-content/uploads/2025/03/IDC_FutureScape_Worldwide_Artificial_Intelligence_and_Automation_2024_Predictions_-_2023_Oct.pdf
www.ey.com/en_in/services/technology/ai-augmented-software-development-a-new-era-of-efficiency-and-innovation
kpmg.com/xx/en/our-insights/transformation/kpmg-global-tech-report-2024.html
www.accenture.com/us-en/insights/consulting/gen-ai-talent
solutionshub.epam.com/blog/post/accelerated-discovery-for-agile-business-analysis
www.it-cisq.org/cisq-files/pdf/CPSQ-2020-report.pdf
