2025 Research

State of AI-Assisted Software Development

The 2025 DORA research is blunt: AI amplifies what is already there. Strong teams ship faster with it; struggling teams get more chaos.

By Burhan Öcüt

DevOps Architect and Engineering Enablement Advisor · Updated 2025-09-23

AI Adoption

90%

of developers use AI at work

Survey Respondents

~5,000

technology professionals

Productivity Boost

80%+

report increased productivity

Trust in AI Code

30%

report little to no trust

Key Takeaway: AI is an Amplifier

The greatest returns on AI investment come not from the tools themselves, but from a strategic focus on the underlying organizational system: the quality of the internal platform, the clarity of workflows, and the alignment of teams. Without this foundation, AI creates localized pockets of productivity that are often lost to downstream chaos.

AI's Impact on Key Outcomes

Effect sizes from the 2025 DORA research. Green = beneficial, Red = needs attention.

Individual Effectiveness

Software Delivery Instability

Organizational Performance

Valuable Work

Code Quality

Product Performance

Software Delivery Throughput

Team Performance

Burnout

Friction

Note: Software Delivery Instability increase is not desirable. Burnout and Friction show no change.

AI Adoption Status

Using AI at Work90%

Not Using AI10%

AI use in software development has become the standard

Trust in AI-Generated Code

A great deal

A lot

20%

Somewhat

46%

A little

23%

Not at all

"Trust but verify" - healthy skepticism is a sign of mature adoption

What Changed from 2024?

Key shifts in AI's relationship with outcomes

Now Positive

Software Delivery Throughput (was negative)
Product Performance (was neutral)
Valuable Work Time (was negative)

Still Positive

Individual Effectiveness
Code Quality
Team Performance
Organizational Performance

Stubbornly Unchanged

Friction (no improvement)
Burnout (no improvement)
Delivery Instability (still increases)

Actionable Insights

AI Adoption is Universal

90% of developers use AI at work, a 14% increase from 2024. The question is no longer 'if' but 'how'.

Focus training on how to critically evaluate and validate AI output, not just encouraging usage.

Instability Persists

AI still increases software delivery instability. The underlying systems haven't evolved to handle AI-accelerated development.

Invest in technical capabilities: automated testing, feature flags, and rapid rollback mechanisms.

Burnout Unchanged

AI doesn't reduce burnout or friction. These are cultural and systemic issues, not tool problems.

Address leadership, priority stability, and generative culture. Technology alone won't fix burnout.

Platform is Prerequisite

Your platform is the strategic prerequisite for getting real value from AI. Low platform quality = negligible AI impact.

Prioritize platform engineering. A poor developer experience hampers your entire AI strategy.

User Focus is Critical

Without user-centric focus, AI adoption can actually HARM team performance.

Keep users' needs as your North Star. AI helps you move fast, so make sure it is in the right direction.

About 1 in 4 Are High Performers

Elite and High Balanced teams together are roughly 24% of teams. Strong, stable delivery at scale is achievable.

Use the team profiles to diagnose where you are and create targeted improvement pathways.

Sources & Citations

All statistics and findings on this page are derived from the 2025 Accelerate State of DevOps Report published by DORA (DevOps Research and Assessment), a program at Google Cloud.

Survey data was collected from approximately 5,000 technology professionals across various industries and company sizes globally.

Report Published

September 2025

Page Last Updated

May 2026

Data Type

Survey Research

Explore Further

Full 2025 DORA Report

Read the complete research

Take Our Assessment

Evaluate your team's maturity

View DevOps Roadmap

Plan your improvement journey

Questions

Frequently asked questions

What are the four DORA metrics?: The four DORA metrics are deployment frequency, lead time for changes, change failure rate, and time to restore service. The first two measure throughput (how fast you ship) and the last two measure stability (how safely you ship), so together they describe delivery performance without trading speed against reliability.
What is a good deployment frequency?: Elite performers deploy on demand, often multiple times per day, while high performers deploy between once per day and once per week. The right target depends on context: the goal is small, frequent, low-risk changes rather than hitting a specific number.
How is change failure rate calculated?: Change failure rate is the share of deployments that cause a production failure needing a hotfix, rollback, or patch. Divide the number of failed deployments by the total deployments over a period. In the DORA research the performance bands are elite 0-5%, high 6-10%, medium 11-15%, and low above 15%.
How is lead time for changes measured?: Lead time for changes is the time from a commit being merged to that change running in production. It reflects how long your pipeline and review process take, and elite performers measure it in hours or less while lower performers measure it in weeks or months.
Do DORA metrics still matter with AI-assisted development?: Yes. The 2025 DORA research found that AI amplifies existing team capability rather than replacing it: strong teams ship faster with AI while struggling teams add more instability. The four metrics remain the way to tell which outcome you are actually getting.

Key Takeaway: AI is an Amplifier

Actionable Insights

AI Adoption is Universal

90% of developers use AI at work, a 14% increase from 2024. The question is no longer 'if' but 'how'.

Focus training on how to critically evaluate and validate AI output, not just encouraging usage.

Instability Persists

AI still increases software delivery instability. The underlying systems haven't evolved to handle AI-accelerated development.

Invest in technical capabilities: automated testing, feature flags, and rapid rollback mechanisms.

Burnout Unchanged

AI doesn't reduce burnout or friction. These are cultural and systemic issues, not tool problems.

Address leadership, priority stability, and generative culture. Technology alone won't fix burnout.

Platform is Prerequisite

Your platform is the strategic prerequisite for getting real value from AI. Low platform quality = negligible AI impact.

Prioritize platform engineering. A poor developer experience hampers your entire AI strategy.

User Focus is Critical

Without user-centric focus, AI adoption can actually HARM team performance.

Keep users' needs as your North Star. AI helps you move fast, so make sure it is in the right direction.

About 1 in 4 Are High Performers

Elite and High Balanced teams together are roughly 24% of teams. Strong, stable delivery at scale is achievable.

Use the team profiles to diagnose where you are and create targeted improvement pathways.

Frequently asked questions

What are the four DORA metrics?

The four DORA metrics are deployment frequency, lead time for changes, change failure rate, and time to restore service. The first two measure throughput (how fast you ship) and the last two measure stability (how safely you ship), so together they describe delivery performance without trading speed against reliability.

What is a good deployment frequency?

Elite performers deploy on demand, often multiple times per day, while high performers deploy between once per day and once per week. The right target depends on context: the goal is small, frequent, low-risk changes rather than hitting a specific number.

How is change failure rate calculated?

Change failure rate is the share of deployments that cause a production failure needing a hotfix, rollback, or patch. Divide the number of failed deployments by the total deployments over a period. In the DORA research the performance bands are elite 0-5%, high 6-10%, medium 11-15%, and low above 15%.

How is lead time for changes measured?

Lead time for changes is the time from a commit being merged to that change running in production. It reflects how long your pipeline and review process take, and elite performers measure it in hours or less while lower performers measure it in weeks or months.

Do DORA metrics still matter with AI-assisted development?

Yes. The 2025 DORA research found that AI amplifies existing team capability rather than replacing it: strong teams ship faster with AI while struggling teams add more instability. The four metrics remain the way to tell which outcome you are actually getting.