Digital.ai 18th State of Agile Report: Key Findings, AI Adoption, and the Future of Software Delivery
Agile software development is entering a new era as artificial intelligence, automation, and enterprise-wide collaboration reshape how organizations plan, build, test, and deliver digital products. Traditional Agile methodologies remain important, but business leaders are increasingly demanding measurable outcomes rather than simply faster development cycles. The Digital.ai 18th State of Agile Report, released on October 28, 2025, examines this transformation and identifies the emergence of what Digital.ai describes as the Fourth Wave of Software Development and Delivery.
The report highlights a significant increase in artificial intelligence adoption, growing pressure to demonstrate the return on investment from Agile practices, and a widening gap between automation and organizational governance. According to the survey, AI adoption increased from 68% to 84%, while only 49% of respondents reported having governance guardrails in place.
These findings suggest that the next stage of Agile transformation involves more than implementing new development tools. Organizations need stronger leadership alignment, reliable performance measurement, responsible AI governance, and better coordination between technology teams and business objectives.
This comprehensive guide examines the report’s major findings, explains the Fourth Wave of software delivery, explores the emergence of Agentic Agile Planning, and outlines practical strategies for organizations seeking to improve software delivery performance.
What Is the 18th State of Agile Report?
The State of Agile Report is a long-running industry research initiative examining how organizations adopt Agile practices, manage software delivery, and respond to changing technological requirements.
Digital.ai published the 18th edition in October 2025, providing insights into Agile adoption, artificial intelligence, organizational transformation, and software delivery performance.
The survey collected responses from nearly 350 participants, primarily Agile coaches and consultants associated with large enterprises employing more than 20,000 people.
This respondent profile is important when interpreting the findings. The survey offers insight into the experiences and perspectives of its participants, particularly those working in large enterprise environments, rather than representing every software development organization worldwide.
Why the Report Matters
Agile practices have influenced software development for more than two decades. However, organizations continue facing difficulties connecting development activities with measurable business outcomes.
The report examines whether companies are improving their ability to respond to change, deliver customer value, and coordinate technology investments.
Its findings are especially relevant to software engineering leaders, Agile coaches, product managers, DevOps teams, and executives responsible for digital transformation.
The Fourth Wave of Software Development and Delivery
One of the central themes of the report is the emergence of the Fourth Wave of software delivery.
Digital.ai uses this concept to describe an industry transition in which artificial intelligence moves beyond supporting individual development tasks and begins participating more actively across the software delivery lifecycle.
Instead of using AI only to generate code or summarize documentation, organizations are exploring systems that can coordinate workflows, identify risks, recommend actions, and execute selected tasks.
From Agile Practices to Enterprise-Wide Intelligence
Traditional Agile frameworks emphasize iterative development, collaboration, feedback, and adaptability.
These principles remain relevant, but modern enterprises often operate large technology portfolios involving numerous development teams, applications, and infrastructure environments.
Coordinating this work requires visibility across planning, development, testing, security, and deployment.
The Fourth Wave envisions AI-supported systems that help connect these activities and provide better information for decisions.
AI as an Active Participant
Earlier AI development tools generally responded to individual prompts and performed limited tasks.
Emerging agentic systems are designed to pursue defined objectives through sequences of actions, potentially using information from multiple tools.
For example, an AI-supported workflow might analyze a software release, identify outstanding testing concerns, summarize delivery risks, and recommend corrective actions.
More advanced systems may execute approved operations when suitable permissions and safeguards are established.
However, the ability to act independently introduces new governance, security, and reliability requirements.
1. AI Adoption Has Increased to 84%
One of the most notable findings is the growth in artificial intelligence adoption among survey respondents.
The Digital.ai 18th State of Agile Report states that adoption increased from 68% to 84%.
This represents a 16-percentage-point increase and indicates that AI tools have become increasingly common within the surveyed organizations.
How Software Teams Are Using AI
Artificial intelligence can support numerous software development activities, including:
- Generating initial code suggestions and reusable components.
- Explaining unfamiliar programming logic.
- Creating draft technical documentation.
- Assisting with software testing and test case generation.
- Identifying potential defects and development risks.
- Summarizing project information and development progress.
- Supporting release planning and operational analysis.
These capabilities can reduce effort for selected tasks, although their effectiveness varies according to application complexity, engineering practices, and the quality of available information.
Coordinated Adoption Across Teams
The report also indicates that 41% of respondents are implementing AI tools in a coordinated manner across teams.
This is an important distinction because individual developers using AI tools does not necessarily mean an organization has established an enterprise AI strategy.
Coordinated implementation can involve shared standards, approved technology platforms, training, security requirements, and consistent performance measurement.
Why AI Adoption Alone Is Not Enough
Higher adoption rates do not automatically demonstrate increased productivity or improved software quality.
Organizations must assess whether AI tools produce meaningful improvements in completed work, system reliability, customer satisfaction, and delivery performance.
Without measurement, AI adoption can become another technology investment whose commercial value remains unclear.
2. Agile Return on Investment Faces Greater Scrutiny
The report identifies increasing pressure to demonstrate the business value of Agile transformation.
According to the findings, 76% of respondents reported greater scrutiny of Agile’s business impact and return on investment.
This reflects a broader shift in how organizations evaluate software development initiatives.
Business leaders increasingly want evidence that Agile practices improve outcomes rather than simply changing internal workflows.
Moving Beyond Activity Metrics
Traditional Agile reporting often includes sprint velocity, completed story points, backlog progress, and team activity.
These measurements can help teams understand their work, but they do not directly prove that customers or businesses receive greater value.
For example, completing more story points does not necessarily mean that a software product is more useful or profitable.
Organizations should connect delivery metrics with customer and business outcomes.
Customer Satisfaction as a Success Measure
The survey reports that customer satisfaction remains a leading Agile success measure, identified by 52% of respondents.
Efficiency and cost reduction follow at 40%.
These findings suggest that organizations increasingly recognize the importance of customer experience and operational value.
Practical Business Outcome Metrics
Depending on the organization, useful measures may include customer retention, product adoption, reduced service interruptions, faster resolution of customer problems, and improved business process efficiency.
Financial measures can include revenue attributable to new capabilities, operating cost reductions, and lower expenses from software defects.
The most appropriate metrics depend on the product, customer base, and business objectives.
3. Organizations Are Being Asked to Do More With Less
Software development teams operate under increasing expectations involving productivity, innovation, and adaptability.
The report identifies several pressures affecting organizations:
- 79% report pressure to accomplish more with fewer resources.
- 78% report pressure to respond quickly to leadership and structural changes.
- 77% report pressure to accelerate innovation.
- 76% report increased scrutiny of Agile business impact and ROI.
These findings help explain why organizations are exploring automation and AI-supported delivery workflows.
Balancing Speed and Quality
Accelerating software releases can provide competitive advantages, but poorly managed speed can increase defects, security vulnerabilities, and operational incidents.
Successful development organizations seek to improve delivery efficiency without sacrificing reliability.
Automated testing, continuous integration, security scanning, and improved deployment processes can help support this balance.
Reducing Repetitive Work
Development teams frequently spend time on repetitive activities such as preparing reports, reviewing routine configuration changes, gathering project updates, and maintaining documentation.
Appropriate automation may reduce this administrative burden.
However, organizations should avoid automating inefficient processes without first understanding and improving the underlying workflow.
4. AI Governance Is Falling Behind Adoption
One of the report’s most important findings concerns the gap between AI adoption and responsible oversight.
Although 84% of respondents reported AI adoption, only 49% indicated that governance guardrails were in place.
This difference suggests that many surveyed organizations are experimenting with AI before establishing comprehensive oversight procedures.
What Are AI Governance Guardrails?
AI governance guardrails are policies, technical controls, and organizational procedures designed to guide the responsible use of artificial intelligence.
They may cover information security, data privacy, approved tools, model evaluation, human oversight, access permissions, and accountability.
For software delivery teams, governance also involves determining which AI-generated actions require human review.
Risks of Inadequate Governance
Weak governance can create several problems.
AI-generated code may contain vulnerabilities or incorrect application logic. Automated systems may misunderstand requirements or act on incomplete information.
Organizations may also expose confidential business information when employees use unapproved external AI services.
These risks become more significant when AI agents can interact with production systems or initiate operational changes.
Creating an Effective Governance Framework
Organizations should establish clear policies defining acceptable AI use, approved tools, data handling requirements, and accountability.
High-impact actions should require appropriate authorization.
Logging, monitoring, evaluation, and incident response procedures should be part of the implementation.
Governance should support responsible innovation rather than unnecessarily preventing useful experimentation.
5. Leadership Alignment Is Essential for Agile Success
The report finds that 73% of respondents want stronger leadership support and clearer alignment between software delivery activities and business objectives.
This finding highlights the importance of organizational coordination.
Even technically capable development teams can struggle when business priorities are unclear or change without effective communication.
Connecting Engineering Work to Strategy
Software development activities should support defined business goals.
For example, a company seeking to improve customer retention may prioritize application reliability, user experience, and customer support integration.
Development teams can then evaluate work according to how it contributes to those objectives.
Improving Portfolio Visibility
Large organizations often operate numerous projects competing for budgets, employees, and infrastructure resources.
Portfolio management helps leadership understand which initiatives are delivering value and which may require adjustment.
AI-supported analytics can assist by summarizing information across projects and identifying potential dependencies.
However, these insights should complement rather than replace informed human decision-making.
Accountability for Outcomes
The survey reports that 29% of respondents are accountable for connecting Agile work with business outcomes.
Another 26% report influencing product and portfolio planning.
These figures indicate that some Agile roles are expanding beyond team-level process support into strategic coordination.
6. Hybrid and Blended Agile Models Are Becoming Common
The report indicates that 74% of respondents use hybrid, blended, or internally developed Agile models.
This suggests that many organizations adapt Agile practices instead of following a single framework without modification.
Why Organizations Customize Agile
Different teams have different requirements.
A small software product team may benefit from short development iterations and frequent customer feedback.
A large enterprise responsible for regulated financial software may require additional documentation, security approvals, and formal release controls.
Agile practices should accommodate these differences while preserving collaboration, adaptability, and continuous improvement.
Combining Agile With DevOps
DevOps practices connect software development with operational responsibilities.
Automated builds, testing, deployment pipelines, monitoring, and incident management can help organizations deliver software more consistently.
Combining Agile planning with DevOps execution may improve coordination between product priorities and operational requirements.
Avoiding Excessive Process Complexity
Customized frameworks can become inefficient when organizations introduce too many meetings, approval stages, or reporting requirements.
Teams should regularly evaluate whether their processes improve outcomes or create unnecessary administrative work.
The objective is effective delivery rather than compliance with a particular process label.
7. Agentic AI Is Emerging in Software Delivery
The Fourth Wave described in the Digital.ai 18th State of Agile Report places particular emphasis on agentic artificial intelligence.
Agentic AI refers to systems designed to pursue objectives through multiple actions, potentially interacting with applications, tools, and information sources.
The report states that more than one-quarter of surveyed AI users are experimenting with agentic systems.
AI Agents in Development Workflows
Potential applications include analyzing project requirements, generating development plans, coordinating testing activities, and identifying software delivery risks.
AI agents may also support certain operational activities, such as summarizing incidents and suggesting remediation steps.
These capabilities remain dependent on system reliability, permissions, and the availability of trustworthy information.
Risk Detection and Predictive Insights
AI-supported analytics can examine historical delivery data and identify patterns associated with delays, defects, or resource constraints.
For example, a system might highlight a project experiencing repeated testing failures or unresolved dependencies.
Such information may help teams investigate problems earlier.
However, predictive outputs are not guarantees and should be validated against real project conditions.
Workflow Orchestration
Agentic systems may coordinate approved actions between project management tools, code repositories, testing platforms, and deployment systems.
For example, an AI-supported workflow could identify a failed test, create a development ticket, and notify the responsible team.
Higher-risk activities, particularly production changes, should remain subject to appropriate safeguards and approvals.
What Is Agentic Agile Planning?
Digital.ai uses the term Agentic Agile Planning to describe an approach in which human teams and AI agents collaborate across planning and software delivery activities.
The model emphasizes using AI to improve information flow, identify risks, and support adaptive decision-making.
Rather than treating project planning as a disconnected administrative activity, the approach aims to connect it more closely with development, testing, security, and operational results.
Continuous Planning
Traditional project plans can become outdated as customer requirements, technical constraints, and organizational priorities change.
AI-supported planning systems may analyze updated information and identify where priorities require review.
Human stakeholders remain responsible for deciding which changes align with business objectives.
Improved Information Visibility
Large development organizations often store information across numerous disconnected systems.
Agentic Agile Planning may help summarize relevant information and reveal relationships between projects, teams, and delivery milestones.
The value of these capabilities depends heavily on data accuracy and integration quality.
Human and AI Collaboration
AI can assist with information processing and routine workflow activities, while human professionals provide context, accountability, creativity, and professional judgment.
This distinction is particularly important when decisions affect customers, security, finances, or operational continuity.
Measuring Agile Performance in the AI Era
The growing emphasis on ROI means organizations need performance meas