Many claims and stories are circulating about AI accelerating development, with promises of compressing a week’s worth of work into just a few hours or an afternoon. In large organizations, however, individual speed rarely translates into the faster delivery of customer value.

For two decades, Agile methods have championed optimizing the entire value stream—identifying bottlenecks, resolving dependencies, and aligning business, technology, and product teams. As AI accelerates individual output, optimizing the system becomes even more critical.

The Barriers To Flow Are Organizational

Slow developers are not your organization’s primary barrier to better speed, agility, and flow. Large organizations have dozens of systemic obstacles that prevent a smooth end-to-end flow. AI agents, by themselves, address only a fraction of these.

AI will not help your teams work better together. AI won’t improve the quality of interactions in your organization. It won’t create purpose, alignment, or ownership, and it won’t build trust.

Organizations embarking on AI transformations, employing AI Agents and AI-enabled teams, will encounter the same frictions that Lean/Agile has been trying to help address. They’ll be stuck navigating cross-organizational dependencies and slowed down by wasteful internal processes. Annual planning cycles will continue to dictate the pace, regardless of how quickly teams can code.

Even if building software becomes lightning-fast, it will increase the demands on all the services surrounding development. Getting your end-to-end value stream to run at an AI-enabled pace will create even more pressure on departments like finance, legal, security, and procurement. Product decisions need to be faster and more frequent. Roadmaps will need to evolve constantly. Finally, fast and reliable test suites will become a necessity, ensuring that AI-accelerated development doesn’t introduce regressions or unintended system behaviours.

AI-Accelerated Development Demands Decentralized Decision-Making

If developers are producing code in minutes, having to wait for managers or committees to approve technical or product decisions will quickly block teams and destroy any productivity gains. To improve flow, power and control will have to shift closer to the work. AI-enabled software teams need significantly more decision-making authority. Moving authority to where the work is being done is not a new idea. Agile has long promoted the ideas of self-organizing teams, peer-level coordination, distributed decision-making, and aligned autonomy.

With AI-driven development, while Agile ideas become increasingly important, many standard practices will need to be reconsidered. Teams will need to reevaluate whether standard Agile practices, such as daily stand-ups or story point estimations, still add value, or if new, leaner ways of working are required. When an AI-enabled developer can develop a new feature in the same time they used to spend in morning stand-ups, the trade-off needs to be reconsidered.

While agility as an organizational capability is more critical than ever, Agile, as it’s commonly practiced, is legacy.

The Real Transformation Happens Within

For teams adopting AI, yes, there will be some new skills to learn, tools to adopt, and costs to manage. But the “heavy lifting” that will make the real difference will come down to the leadership, culture, structures and systems within your organization.

Don’t look externally, expecting AI to transform your organization. The real difference comes from what happens inside your company.

The problems holding teams back—low-trust environments, old management styles, controlling cultures—won’t vanish with AI. Speed, agility, and flow are never problems about technology or methodology; they arise from a company’s culture and its systems.

With AI-enabled software teams, close collaboration between business, product, and technology remains vital. Strong product management capabilities, guiding the vision and understanding user needs, become the real difference maker.

With new tools and new potential, developers will quickly get frustrated if they’re constantly stuck in bureaucratic project control meetings and forced to do waterfall planning. Talented developers who want to use AI to build meaningful products and solutions may leave for better opportunities.

I expect the majority of AI transformations will follow a similar path to many Agile transformations. Companies that viewed Agile as a set of practices to implement for faster development and better quality, but that didn’t make the necessary cultural and systematic changes, failed to achieve meaningful results. Agility was always an outcome of mature leadership, culture and systems. Making these same changes will be even more challenging in the era of AI.