Agile changed the world of work. Now AI is changing the world that Agile was designed for. The next revolution may require us to challenge the very thinking that started the last one.
In 2001, seventeen software developers challenged the status quo. Their Agile Manifesto was a declaration of independence from rigid processes, heavyweight documentation and the belief that successful software could be delivered through planning alone.
It was radical. It was influential. And it helped change how organisations approached work.
But a quarter of a century later, the world has changed dramatically. Artificial intelligence is transforming how we think, create, collaborate and deliver. Machines can generate code, analyse information, produce designs and execute increasingly complex tasks at a speed that would have seemed extraordinary when the manifesto was written.
Yet much of the Agile industry remains attached to a model conceived for a different era.
What if Agile's greatest strength—its willingness to challenge established thinking—has become its greatest blind spot when it comes to challenging itself?
Perhaps it is time for an Agile bonfire.
Not to erase the past. Not to discard everything that still works. But to ask a question the Agile community should be brave enough to confront:
If we were inventing Agile today, knowing what we know about AI, would we invent the same thing?
I suspect not.
The World Has Changed. Has Agile?
The original manifesto emerged from a world in which human teams did most of the thinking, designing, coding, testing and coordinating required to deliver software.
Its values reflected that reality. People and interactions mattered more than processes and tools. Working software mattered more than comprehensive documentation. Collaboration mattered more than contractual negotiation. Responding to change mattered more than following a plan.
The 2001 Agile Manifesto remains the essential starting point for that conversation. Its four values still prioritise people, working software, customer collaboration and responsiveness to change. Read the original manifesto.
These principles remain powerful. As explored in The Agile Mindset, adaptability is an organisational capability, not simply a delivery technique.
But AI introduces a fundamental shift. It changes not merely how quickly people perform existing tasks, but who—or what—performs them, how work is organised and where the constraints on performance actually lie.
Consider the implications.
AI can generate and test code, create prototypes, analyse customer feedback and explore multiple solutions simultaneously. AI agents can increasingly undertake tasks, use tools and coordinate sequences of work with varying degrees of human supervision.
The limiting factor is no longer always the speed at which a human team can produce something. It may be the quality of the question, the clarity of the objective, the organisation's ability to make decisions, the reliability of its information or the time it takes to approve a change.
When the economics of creating work change, the economics of organising work must change too.
And that is where the real challenge to Agile begins.
What the Evidence Actually Says About AI and Agile
AI adoption is not the same as organisational transformation. A 2026 Scrum.org survey of 289 Agile practitioners found that 83% of respondents used AI tools, but 55% spent 10% or less of their working time using them. Only 15% reported receiving formal training in AI for Agile work. The sample is limited and should not be treated as representative of every Agile organisation, but it illustrates the gap between access to AI and embedding it meaningfully in practice.
The real question is therefore not whether teams have acquired AI assistants. It is whether those tools have changed how they discover problems, validate solutions and make decisions.
Source: Scrum.org, The AI4Agile Practitioners Report 2026
When Yesterday's Solution Becomes Tomorrow's Constraint
Agile was never synonymous with Scrum, two-week sprints, story points, stand-ups or backlogs. The original manifesto did not prescribe those things as universal requirements.
Yet an industry has grown around particular frameworks, ceremonies, roles, metrics, certifications and interpretations of what Agile should look like.
For many organisations, the result is a curious contradiction: a movement founded on adaptability has acquired its own machinery of conformity. A related critique in Agile: Treat or Trick? asks whether following Agile practices has become a substitute for achieving genuine business agility.
AI exposes that contradiction.
If a team can generate and test a working solution in hours, does every piece of work need to follow the same planning and delivery rhythm?
If AI can automate routine coordination and reporting, why preserve meetings and administrative processes simply because they are embedded in the framework?
If several AI agents can explore alternative solutions in parallel, should the organisation still structure every decision around a sequential backlog and a series of human handoffs?
Sometimes the answer will be yes. Dependencies, risk, uncertainty and the nature of the work still matter. Disciplined planning and regular feedback can remain valuable.
But the answer should emerge from the needs of the work, not from unquestioned loyalty to a method.
The danger is not that Agile has too much discipline. It is that organisations confuse the discipline required to achieve an outcome with the rituals historically used to coordinate one.
Automating an outdated operating model does not create agility. It accelerates its limitations.
A Practical Example: The Two-Week Sprint Meets AI
Imagine a product team that needs to test whether customers understand a new registration journey. In a conventional delivery cycle, the team might discuss requirements, write user stories, estimate effort, schedule design and development, and wait until the next sprint review to gather feedback.
With AI assistance, the same team might generate several prototype variants and draft test scripts in a single day. Researchers could then put those prototypes in front of real users, evaluate confusion points and decide which idea deserves production engineering. These are illustrative possibilities, not a promise that AI will complete the work accurately or safely.
The vital work has not disappeared: protecting personal information, checking accessibility, assessing security, interpreting user feedback and deciding what is worth building. What changes is the opportunity to shorten the cycle between hypothesis and evidence. If the team continues to wait two weeks to make a decision that could be made tomorrow, the limiting factor may be governance rather than coding.
Conversely, where software is safety-critical, highly regulated or deeply interconnected, slower review and explicit approval may be essential. True agility means choosing the right cadence for the risk, not universally removing controls.
From Agile Teams to Intelligent Organisations
The next evolution must go beyond making existing teams more productive.
It must rethink the organisation itself.
Traditional Agile places considerable emphasis on how teams organise, collaborate and deliver. Those capabilities still matter. But in an AI-enabled organisation, agility increasingly depends on how the entire system senses change, interprets information, makes decisions, allocates resources and acts.
A highly productive team cannot compensate for a leadership team that takes six months to approve an experiment. Faster software development will not rescue a business that cannot recognise a changing customer need. An AI-powered workforce will not create meaningful advantage if every decision still has to travel through layers of bureaucracy. That gap between formal structures and real delivery is examined in The Gap Between How Organisations Are Designed to Work—and How They Actually Deliver.
The bottleneck may have moved.
The question is no longer simply how quickly teams can deliver work. It is how quickly an organisation can turn new information into better decisions and better outcomes.
That requires a different balance of capabilities.
Humans bring judgement, imagination, empathy, context, ethical reasoning and the ability to question whether the right problem is being solved. AI can extend our ability to analyse, generate, simulate and execute.
Neither is sufficient alone.
AI can produce an answer without understanding whether the question matters. It can optimise a process whose purpose is no longer relevant. It can generate impressive volumes of output without creating corresponding value.
The competitive advantage will not necessarily belong to the organisation with the most AI, the most Agile teams or the fastest delivery pipeline.
It will belong to the organisation that combines human judgement and artificial intelligence most effectively to learn and adapt.
This direction is already being discussed beyond individual teams. In March 2026, PMI Agile Alliance published a Manifesto for Enterprise Agility, highlighting the need to connect organisational adaptability with leadership, decision-making and realised value. That development strengthens the case for widening the discussion, even if reasonable practitioners disagree about how far existing Agile principles need to be rewritten.
Further reading: PMI Agile Alliance, Manifesto for Enterprise Agility (2026)
The Adaptive Intelligence Manifesto
If the Agile Manifesto helped redefine software development, perhaps the next manifesto should redefine how organisations create value in an age of accelerating intelligence.
Call it the Adaptive Intelligence Manifesto.
Not a finished standard or another framework to impose on everyone, but a starting point for a necessary debate.
We are discovering better ways of creating value in a world of accelerating intelligence and change. We value:
- Purpose over activity. Meaningful outcomes matter more than the volume of work produced.
- Adaptive intelligence over fixed methods. We change how we work to fit the challenge, rather than forcing every challenge into the same process.
- Continuous discovery over assumed certainty. We test, learn and revise our understanding before committing too heavily to a solution.
- Human judgement and machine intelligence over either alone. We combine their strengths without confusing computational capability with wisdom.
- Flow of value over flow of tasks. We optimise the whole system, not local utilisation, ceremony completion or individual productivity.
- Responsible autonomy over centralised control. We distribute authority while making accountability, safety and boundaries explicit.
- Continuous adaptation over periodic adjustment. We sense change, act on evidence and learn at the pace the situation demands.
These proposed values are not a repudiation of the original manifesto. They build on its spirit while shifting the emphasis towards the challenges emerging now.
The central change is profound: agility becomes less about following a preferred delivery methodology and more about designing a system capable of continually adapting how it thinks, decides and acts.
The goal is not to make every organisation operate at machine speed. It is to remove unnecessary constraints while preserving the judgement, safeguards and human relationships that make speed worthwhile.
The Leadership Problem Nobody Can Automate Away
There is something uncomfortable behind the AI revolution.
Many organisations will invest heavily in AI while leaving their underlying operating models largely untouched.
They will introduce AI assistants, automate reporting, generate more code and accelerate individual tasks. They may even measure success through spectacular increases in output.
But if priorities remain confused, decision-making remains slow, accountability remains fragmented and success is still measured by activity rather than outcomes, the fundamental problem will survive. Possibly in a more expensive form.
Leadership must therefore move beyond asking how AI can improve the existing Agile process. It must ask what should change about the process, the structure and the organisation itself.
That means challenging unnecessary approvals. Rethinking decision rights. Designing effective collaboration between people and AI. Establishing proportionate controls for automated decisions. Measuring realised value rather than merely counting what has been produced. The point echoes Agile Needs Champions: sustainable change requires people willing to challenge entrenched decisions, not merely enforce a framework.
Above all, it means accepting responsibility for redesigning the system rather than blaming teams for failing to deliver faster within it.
AI may accelerate execution. Only leadership can decide whether the organisation is executing the right things.
That is the accountability test for the next era of agility.
How Leaders Can Adapt Agile for the AI Era
- Audit the actual bottlenecks. Follow one idea from customer need to measurable outcome. Identify where delays arise from production, handoffs, approval queues, missing evidence or unclear ownership.
- Separate experimentation from high-risk release. Encourage fast prototyping while keeping appropriate human review, privacy protection, security testing and deployment controls.
- Give teams bounded decision authority. Set clear objectives, budgets, escalation thresholds and accountability so routine choices do not need to travel through unnecessary layers.
- Measure outcomes, not AI-generated volume. Look at validated customer improvements, time to learning, quality, reliability and value delivered—not just prompts written, tickets closed or lines of code produced.
- Revisit the operating model regularly. Ask which meetings, roles and workflows genuinely improve coordination and which survive because the organisation has stopped questioning them. To make those reviews tangible, leaders can use the Team Health Check and Meeting Cost Calculator to surface team friction and the cost of recurring coordination.
This is consistent with a broader argument already explored on Agile.org.uk: culture and learning matter more than mechanical adherence to a framework.
Related reading on Agile.org.uk: Agility That Works: Culture First, Framework Second
Burn the Dogma. Keep the Courage.
Should we really burn the Agile Manifesto?
Taken literally, no. Its core values have not suddenly expired. Responding to change, collaborating with customers and valuing people remain as relevant as ever.
But treating the manifesto as untouchable would betray the very spirit that made it influential.
Its greatest contribution was not a permanent set of instructions. It was the courage to question the accepted wisdom of its time. That courage must now be applied to Agile itself.
We should preserve what remains valuable, challenge the practices that have become counterproductive and create new principles where the world demands them.
We should stop treating Agile as a destination and start treating agility as a capability that must continually evolve.
And we should resist the temptation to turn the next manifesto into another rigid doctrine, complete with its own ceremonies, certifications and commercial machinery.
That would simply repeat the cycle.
The bonfire is not an invitation to destroy everything that came before. It is an invitation to distinguish enduring principles from inherited habits.
Because the ultimate failure would not be abandoning Agile.
It would be preaching adaptation while refusing to adapt ourselves.
Agile changed how we deliver. AI challenges us to rethink how we work. The next revolution will belong to those prepared to question both.
Perhaps the time has come to light the bonfire—not to bury Agile, but to set its future free.
