AI Adoption & Impact Study: AI in Software Development June 26 Top 10 Insights

For enterprise IT organizations, AI is more than a rapidly evolving technology; it represents a foundational shift across the software development landscape. Our AI in Software Development June 2026 Top 10 Insights report paints a detailed portrait of AI adoption in the software development lifecycle (SDLC). Key insights include AI’s impact on code quality and developer productivity, plus challenges ranging from security concerns to AI adoption barriers within development workflows.

As part of our ongoing AI Adoption and Impact Study, over 500 technical practitioners responded to our AI in Software Development survey. The resulting report offers timely, evidence-based data on AI coding tools, security practices, and adoption maturity in the software development process. This research sheds light on where AI is bringing significant improvements to software development and where obstacles to AI integration and value delivery remain within the SDLC.

10 key insights on the state of AI in the SDLC today.

These critical findings examine current AI adoption practices in software development, the gains achieved from them, and where AI-assisted outcomes are falling short of expectations.

1. AI speeds up code creation, but the code created requires more review.

Although 84% of developers use AI in the Build phase of software development, two-thirds of developers say AI-generated code requires more testing than human-generated code.

2. More experienced practitioners are significantly more likely to apply AI security tools.

Practitioners with 15+ years of experience are nearly twice as likely to use AI security tools as those with just 3-7 years of experience. The most experienced practitioners are also four times more likely to raise security or codebase context concerns in open text responses.

3. Engineers and applications teams disagree with product teams that vibe coding is error-free.

While 41% of product team members agree that vibe coding guarantees “zero errors,” 51% of applications practitioners disagree with that. Engineers have a more neutral opinion on the matter. Engineering and applications professionals need a formal seat at the table when AI adoption decisions are made, not just when the quality problems surface afterward.

4. AI tool adoption outpaces maturity at the Build stage.

Only 37% of developers using AI at the Build stage describe their AI maturity level as “formal or better,” indicating a gap between adoption and process maturity.

5. AI adoption is hindered by security, output quality, and data quality concerns, particularly at medium-to-large organizations.

Output quality is the top AI adoption challenge reported by small organizations (64%), but security concerns dominate for medium (47%) and large (51%) organizations.

6. Working with legacy code is the biggest roadblock to using AI in a given workflow.

When asked to name personal roadblocks to using AI in workflows, 51% of respondents cited AI breaks on legacy code, followed by AI code not passing quality gates (46%) and inconsistent skill levels across the team (36.7%).

7. Almost all developers report that their productivity has improved. Higher AI maturity correlates with further productivity gains.

Over 90% of developers report meaningful AI productivity gains. Organizations that report higher AI maturity at the Build phase are also more likely to report higher developer productivity.

8. AI is reducing defects in shipped code for most practitioners; more than half report defects have dropped by 50% or more.

While over 80% report meaningful AI defect reduction, only 9% find “no meaningful change,” and 8% claim “it’s too early to measure.” Teams that do not experience a reduction in defects from AI often face a measurement gap rather than an AI adoption problem.

9. Senior developers are more likely to shift time to review AI code, while juniors are more likely to see speed gains.

Although 38% of veteran developers say AI allows them to shift to reviewing and validating code, only 20% of the least experienced developers see that as a benefit of AI coding. Similarly, 75% of the least experienced developers see speed and efficiency gains as a key benefit of AI coding, but only 55% of veteran developers do.

10. More formal AI processes reduce defects in code.

Organizations with an informal level of AI maturity are the least likely to report that AI helps them reduce code defects.