Harness agentic AI in software development

The conversation about AI in software development has shifted from "should we adopt it?" to "how do we manage what we've already deployed?"

For the second year in a row, we worked with KPMG and CIO Dive on a survey to uncover how AI is impacting the software development lifecycle as agentic AI gains traction. We are seeing thatAI adoption has reached a tipping point, with 99% of organizations now incorporating AI into their software development lifecycle (SDLC) processes, which is up from 97% just one year ago.

Of that 99%, 93% report measurable results from their AI implementations. The IT leaders surveyed want to make it clear that these are production systems that are delivering quantifiable value across every stage of software development. This is changing application development as we know it, but in ways that could be considered surprising.

AI transforms the entire development spectrum

The data shows that AI is used for testing and quality assurance, which is to be expected, but it doesn’t stop there. AI can now be found in the more strategic areas of the software development lifecycle (SDLC), such as design and architecture, deployment and integration, and requirements gathering and planning. These are all areas that were largely manual just two years ago.

The most popular emerging use cases reflect where organizations see the greatest business value. Two out of five executives have already adopted AI for code generation, code quality reviews, and automated testing. Perhaps more significantly, 44% report that they are ready to adopt AI for application generation, suggesting significant near-term growth potential.

This expansion into strategic areas indicates that AI adoption has matured. Organizations that started with pilots are now seeing efficiency gains in code generation and application testing, which gave them the confidence to expand AI usage across broader SDLC processes.

Business impact drives continued investment

The measurable results organizations are experiencing explain why AI investment continues at high levels, with 94% planning to increase their AI spending over the next 12 months. The reported impacts speak directly to core business priorities:

Developer productivity gains are substantial

Of the organizations surveyed, 69% report that AI adoption has increased developer productivity because it can do routine, labor-intensive tasks while developers increase their involvement in strategic, business-aligned activities. This elevates the perception and role of development teams in your organization.

Software quality improvements are measurable

When it comes to improving software quality and reducing bugs, 68% of organizations report that AI can accelerate delivery without sacrificing reliability. This quality improvement is critical for building organizational trust in AI-driven development processes.

Scalability becomes achievable

According to 62% of the survey respondents, AI has made development efforts more scalable, enabling faster time to market and more responsive alignment with business needs. This scalability advantage becomes particularly important as organizations face increasing pressure to deliver software capabilities at enterprise scale.

AI sprawl emerges as the next critical challenge

The success of AI adoption has created an unexpected problem. Too many AI tools are creating fragmentation across development organizations. For larger organizations, the more the use ungoverned generative AI to generate code or applications, the greater the risk of creating problems across the IT landscape.

This AI sprawl challenge extends beyond technical concerns to fundamental issues of security, governance, and compliance. Unapproved AI tools are on the rise, mirroring the shadow IT challenges of the past, but with potentially greater security and compliance implications.

Industry experts emphasize that fragmentation can only take organizations so far before requiring rationalization and consolidation. Success with AI-driven development require platform approaches to AI implementation.

Platform strategies emerge as the solution

The response to AI sprawl challenges is driving organizations toward more strategic platform approaches. Ninety-three percent of organizations are developing or planning to develop their own custom AI agents, using combinations of traditional coding, open-source frameworks, and low-code tools. This high percentage indicates a clear appetite for consolidation and standardization in AI adoption.

Platform approaches offer several advantages for managing AI at scale. These include tool consolidation, reduced redundancy, and maintained governance across the complete AI lifecycle. This enables organizations to scale AI implementations more efficiently and securely while maintaining the control necessary for enterprise environments.

The platform trend also reflects broader organizational learning about technology adoption. Organizations that experienced challenges with proliferated cloud services, development tools, and SaaS applications are applying those lessons to AI implementations, seeking integrated approaches from the beginning rather than trying to rationalize diverse tools after the fact.

Strategic implications for development organizations

Our AI research reveals several strategic imperatives for organizations managing AI adoption across their development lifecycle.

Focus on governance early

The AI sprawl challenge demonstrates that successful AI adoption requires governance frameworks from the beginning, not as an afterthought. Organizations that establish clear policies for AI tool selection, usage, and integration will avoid the technical debt and security risks that come with unmanaged AI proliferation.

Prioritize platform approaches

The trend toward custom AI agent development and platform consolidation suggests that sustainable AI strategies require integrated approaches. Organizations should evaluate their AI tool portfolios and develop consolidation strategies that maintain innovation while reducing complexity.

Measure business impact consistently

The high percentage of organizations reporting measurable AI impact indicates that successful implementations include clear metrics and regular assessment. Development organizations should establish baseline measurements and track specific business outcomes rather than just technical metrics.

The path forward: Managing AI at enterprise scale

The data makes it clear that AI has become a fundamental component of software development operations. The challenge now is managing this transformation at enterprise scale while maintaining the governance, security, and quality standards that enterprise software requires.

Organizations that succeed in this next phase will be those that can balance AI innovation with operational discipline and can scale AI across their development lifecycle without losing control of their technology landscape. That requires not just better AI tools, but better approaches to managing AI as a strategic enterprise capability.

For more insights, read the full report: Navigating Agentic and Generative AI in Software Development.