State of AI-assisted Software Development, 2025 (Google Cloud / DORA, v.2025.2, 142 pp.)

With a high degree of certainty, when organizations have a clear and communicated AI stance, AI's positive influence on individual effectiveness and on organizational performance is amplified, and AI's neutral effect on friction is made beneficial (friction decreases).

so what: The report says A clear AI policy boosts individual output, org results, and cuts friction — all with high certainty.. The source only says A clear AI policy amplifies individual output and org results with high certainty; the friction benefit is found with lesser certainty..

report: §Clear and communicated AI stance p52

reason: The passages confirm high-certainty amplification of individual effectiveness and organizational performance from a clear AI stance. However, the friction finding is explicitly stated at only a "lesser degree of certainty," not a high degree — the claim overstates the confidence level for that specific finding.

With a high degree of certainty, we found that AI adoption's positive benefits depend on organizations having a clear and communicated AI stance, such that, when they do: — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.52

With a lesser degree of certainty, we also found that, in the presence of a clear and communicated AI stance: 1. AI's positive influence on software delivery throughput is amplified. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.52

A clear and communicated AI stance determines AI's impact on organizational performance — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.51

Establishing and socializing a clear policy on permitted AI tools and usage builds developer trust and provides the psychological safety needed for effective experimentation.

report: §Clarify and socialize your AI policies p63

reason: Passage report#p64#1 states nearly verbatim that establishing and socializing a clear policy on permitted tools builds developer trust and provides psychological safety for effective experimentation, matching the claim's subject, scope, and direction exactly.

Ambiguity around AI stifles adoption and creates risk. Establish and socialize a clear policy on permitted tools and usage to build developer trust. This clarity provides the psychological safety needed for effective experimentation, reducing friction and amplifying AI's positive impact on individual effectiveness and organizational performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.64

With a high degree of certainty, AI adoption's positive benefits depend on organizations having healthy data ecosystems, such that AI's positive influence on organizational performance is amplified.

report: §Healthy data ecosystems p54

reason: Passage report#p54#2 confirms that high-quality, accessible, unified data ecosystems yield higher organizational performance benefits than AI adoption alone. Passage report#p55#1 and report#p52#5 confirm the "high degree of certainty" framing for related capabilities. The claim matches the source's substance and confidence level.

The benefits of AI on organizational performance are significantly amplified by a healthy data ecosystem. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.64

With a high degree of certainty, we found that AI adoption's positive benefits depend on organizations having AI-accessible internal data — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.55

With a high degree of certainty, when organizations have AI-accessible internal data, AI's positive influence on individual effectiveness and on code quality is amplified.

so what: The report says having internal data accessible to AI boosts both personal output and code quality. The source only says internal data access amplifies AI's benefit for personal output, and separately for code quality.

report: §AI-accessible internal data p55

reason: Passage report#p52#5 confirms individual effectiveness and code quality are both amplified, and report#p55#3 confirms the code quality finding specifically. However, the individual effectiveness finding is tied to AI-accessible internal data (report#p55#4–6), while report#p54#1 ties organizational performance (not individual effectiveness) to healthy data ecosystems. The claim merges two findings that the source treats as distinct capabilities: AI-accessible internal data amplifies individual effectiveness AND code quality — both confirmed with high certainty per report#p52#5 and the section headers in report#p55#4.

1. AI's positive influence on individual effectiveness is amplified; 2. AI's positive influence on reported organizational performance is amplified; and 3. AI's neutral effect on friction is made beneficial and shown to decrease friction. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.52

2. AI's positive influence on code quality is amplified. While AI tools trained on a general set of knowledge help developers feel more effective and produce higher-quality code, this finding suggests that AI can be even more impactful toward those goals when given access to internal data sources that allow developers to provide their AI tools with company-specific context. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.55

With a high degree of certainty, we found that AI adoption's positive benefits depend on organizations having healthy data ecosystems, such that, when they do, AI's positive influence on organizational performance is amplified. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.54

With a high degree of certainty, AI adoption's positive benefits depend on frequency of version control commits (amplifying AI's influence on individual effectiveness) and on use of rollback/revert features (amplifying AI's influence on team performance).

so what: The report says both version control commit frequency and rollback use are proven with high certainty to boost AI's benefits. No held source says this.

report: §Strong version control practices p56

reason: The passages confirm rollback amplifies AI's team performance benefit (report#p56#6) and commit frequency relates to individual effectiveness (report#p57#2–4), but only the small-batches capability explicitly carries "high degree of certainty" language (report#p58#1). The claim asserts high certainty for both version-control sub-capabilities; the passages do not state that level of confidence for commit frequency specifically.

With a high degree of certainty, when teams work in small batches, AI's positive influence on product performance is amplified and its neutral effect on friction is made beneficial; conversely, AI's benefits for individual effectiveness are slightly reduced.

report: §Working in small batches p58

reason: All three sub-claims are directly supported: amplified product performance (p58#3), friction made beneficial (p58#6), and individual effectiveness slightly reduced (p58#2), all stated with high certainty.

With a high degree of certainty, we found that AI adoption's positive benefits depend on teams working in small batches, such that, when they do: — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.58

1. AI's positive influence on product performance is amplified; and — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.58

In this case, working in small batches increases reported product performance, while also decreasing perceived friction for AI-assisted teams. We think these benefits outweigh any potential harm to individual effectiveness from working in small batches — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.58

For teams who prioritize working in small batches, it seems natural that observed gains in individual effectiveness would be somewhat less. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.58

With a high degree of certainty, on teams with a user-centric focus AI's positive influence on team performance is amplified; in the absence of a user-centric focus, AI adoption has a negative impact on team performance.

report: §User-centric focus p60

reason: Passages report#p95#5 and report#p60#2 state with high certainty that a user-centric focus amplifies AI's positive influence on team performance, and that without it AI adoption can have a negative impact on team performance — matching the claim exactly.

We found with a high degree of certainty that when teams adopt a user-centric focus, the positive influence of AI on their performance is amplified. Conversely, in the absence of a user-centric focus, AI adoption can have a negative impact on team performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.95

they receive an even greater benefit from AI when they center their users, and experience negative impacts from AI adoption when they do not. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.60

DORA found with a high degree of certainty that when teams adopt a user-centric focus the positive influence of AI on their performance is amplified, and that in the absence of a user-centric focus AI adoption can have a negative impact on team performance.

report: §Focus on the user p95

reason: Passages report#p60#1 and report#p60#2 state with high certainty that user-centric focus amplifies AI's positive influence on team performance, and that absence of it causes AI adoption to have a negative impact — matching the claim precisely.

With a high degree of certainty, we found that AI adoption's impacts depend on teams having a user-centric focus. Specifically, when used on teams that adopt a user-centric focus, AI's positive influence on reported team performance is amplified. Importantly, we also found that, in the absence of a user-centric focus, AI adoption has a negative impact on team performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.60

With a high degree of certainty, when organizations have quality internal platforms, AI's positive influence on organizational performance is amplified; conversely, AI's neutral effect on friction is made harmful (more friction).

so what: The report says good platforms boost AI's gains on performance AND make friction worse. The source only says good platforms boost AI's gains on performance; the friction finding is about small batches, not platforms.

report: §Quality internal platforms p62

reason: The passages confirm that quality internal platforms amplify AI's positive influence on organizational performance. However, the "neutral effect on friction made beneficial" finding is attributed to working in small batches (report#p52#5, report#p58#5), not to quality internal platforms. The claim incorrectly attributes the friction moderation — and inverts its direction (harmful vs. beneficial) — to platforms.

with quality internal platforms, AI's positive influence on organizational performance is amplified. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.62

1. AI's positive influence on individual effectiveness is amplified; 2. AI's positive influence on reported organizational performance is amplified; and 3. AI's neutral effect on friction is made beneficial and shown to decrease friction. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.52

With a high degree of certainty, we found that AI adoption's positive benefits depend on organizations having a clear and communicated AI stance, such that, when they do: — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.52

AI adoption has a negligible effect on organizational performance when platform quality is low, but when platform quality is high the effect is strong and positive.

so what: The report says AI barely helps performance with a weak platform, but strongly boosts it with a good one. The source only says A high-quality platform amplifies AI's positive influence on how well the whole company performs.

report: §The strategic imperative p71

reason: The passages confirm a high-quality platform amplifies AI's positive impact on organizational performance. However, no passage states the effect is "negligible" when platform quality is low — only that the amplification depends on platform quality. The "negligible when low" half is not supported.

with quality internal platforms, AI's positive influence on organizational performance is amplified. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.62

A high-quality platform amplifies the effects of AI adoption on organizational performance. The positive impact of AI on organizational performance is strong when platform quality is high. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.66

A high-quality platform serves two purposes when amplifying the impacts of AI on organizational performance. First, it acts as the distribution and governance layer required to scale the benefits of AI from individual productivity gains to systemic organizational improvements. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.72

An investment in AI without a corresponding investment in high-quality platforms is unlikely to yield significant returns at the organizational level. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.72

A high-quality internal platform has a broad, statistically positive effect across outcomes (organizational performance, product performance, productivity), together with a small but credible increase in software delivery instability.

report: §A force multiplier for performance, well-being, and risk p70

reason: Multiple passages confirm quality platforms broadly improve performance outcomes and are associated with a small but credible increase in software delivery instability, matching the claim's subject, scope, and direction.

Consistent with past research, we found that a better platform is associated with a small but credible increase in software delivery instability, meaning a higher change failure rate and increased rework. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.70

The slight increase in instability should be seen as a manageable trade-off for the significant gains in performance that the platform enables. The improvements in product performance are likely more impactful than the modest gains in delivery throughput and reduction in delivery stability. Investing in a high-quality platform is a powerful strategic lever with widespread returns. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.70

High-quality platforms are a force multiplier, improving organizational performance, productivity, and team well-being. A platform should be seen as a holistic entity that enables a great developer experience. A platform functions as an engine for managing risk, enabling speed and experimentation that corresponds to a small but credible increase in software delivery instability: a manageable tradeoff for higher performance overall. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.66

Estimated effect of quality internal platforms on key outcomes Estimated effect of quality internal platforms on key outcomes Individual effectiveness Organizational performance Product performance Code quality Team performance — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.70

The slight increase in instability should be seen as a manageable trade-off for the significant gains in performance the platform enables; investing in a high-quality platform is a powerful strategic lever with widespread returns.

report: §A force multiplier for performance, well-being, and risk p70

reason: Multiple passages directly state the instability increase is small and manageable, offset by broad performance gains, and that high-quality platforms are a force multiplier with wide-ranging positive impact — matching both elements of the claim.

A platform functions as an engine for managing risk, enabling speed and experimentation that corresponds to a small but credible increase in software delivery instability: a manageable tradeoff for higher performance overall. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.66

A great platform acts as a force multiplier, which translates directly into better performance and productivity. The corresponding increase in instability may be representative of a healthy high-velocity system, where the additional instability is acceptable as long as it doesn't impact product performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.70

A high-quality platform, as defined by our capabilities, has a broad, statistically positive impact across the board. It's linked to higher organizational performance, product performance, and productivity. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.70

High-quality platforms are a force multiplier, improving organizational performance, productivity, and team well-being. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.66

While AI adoption on its own shows a modest impact, the effect is dramatically amplified in organizations with strong VSM practices.

so what: The report says AI alone has little effect, but VSM makes that effect dramatically bigger. The source only says VSM acts as a force multiplier, turning AI gains into broader organizational advantages.

report: §Conclusion p78

reason: The source confirms VSM amplifies AI's organizational impact, but never describes AI's standalone effect as "modest" nor the amplification as "dramatic." The claim overstates both the baseline weakness and the magnitude of the VSM boost.

Value stream management (VSM), the practice of visualizing, analyzing, and improving the flow of work from idea to customer, acts as a force multiplier for AI, ensuring that local productivity gains translate into measurable improvements in team and product performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.4

VSM turns AI into an organizational advantage: We hypothesize that VSM moderates the relationship between AI adoption and organizational performance. Teams with mature VSM practices can channel the productivity gains from AI toward solving system-level problems, ensuring that individual improvements translate into broader organizational success. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.77

VSM acts as a force multiplier for AI investments. By providing a systems-level view, it ensures AI is applied to the right problems, turning localized productivity gains into significant organizational advantages instead of simply creating more downstream chaos. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.5

VSM drives team performance: teams that consistently review and improve their value stream report markedly higher performance.

so what: The report says Teams that keep reviewing their value stream score much higher on team results. The source only says VSM practices lift team and product output, and act as a force booster for AI gains at the org level.

report: §How this appears in our 2025 findings p77

reason: The passages confirm VSM has a "direct and powerful impact on performance" and amplifies AI's organizational impact, but the specific claim that teams "consistently reviewing" their value stream report "markedly higher team performance" is not directly stated — the passages focus on VSM's impact on organizational performance and AI amplification, not team performance specifically as a measured outcome of iterative VSM review.

Our findings confirm that organizations that embrace the principles of VSM see significant, measurable benefits. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.77

we found that teams who focus on understanding their value streams dedicate significantly more of their time to valuable work. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.74

Value stream management (VSM), the practice of visualizing, analyzing, and improving the flow of work from idea to customer, acts as a force multiplier for AI, ensuring that local productivity gains translate into measurable improvements in team and product performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.4

While AI adoption on its own shows a modest impact, the effect is dramatically amplified in organizations with strong VSM practices. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.78

VSM leads to more valuable work: teams that understand their value stream spend significantly more of their time on work that matters to the organization and its customers.

report: §How this appears in our 2025 findings p77

reason: Multiple passages directly confirm that teams understanding their value stream dedicate significantly more time to valuable work, matching the claim's subject, direction, and scope without overstatement.

we found that teams who focus on understanding their value streams dedicate significantly more of their time to valuable work. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.74

Teams that work together to understand their value stream spend more time on work that matters. When teams share a clear understanding of their entire value stream, they can focus their efforts on what matters most, translating that clarity into meaningful impact. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.77

VSM improves product performance: a focus on the value stream translates into better product outcomes.

so what: The report says Focusing on the value stream leads to better product results. The source only says VSM raises team and overall company results, acting as a booster for AI gains.

report: §How this appears in our 2025 findings p77

reason: The passages confirm VSM drives higher team and organizational performance and acts as a force multiplier for AI. No passage specifically states VSM improves product performance as a distinct outcome; the claim's scope (product performance) is narrower than what the source supports.

Value stream management (VSM), the practice of visualizing, analyzing, and improving the flow of work from idea to customer, acts as a force multiplier for AI, ensuring that local productivity gains translate into measurable improvements in team and product performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.4

VSM acts as a force multiplier for AI investments. By providing a systems-level view, it ensures AI is applied to the right problems, turning localized productivity gains into significant organizational advantages instead of simply creating more downstream chaos. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.5

Our findings confirm that organizations that embrace the principles of VSM see significant, measurable benefits. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.77

While AI adoption on its own shows a modest impact, the effect is dramatically amplified in organizations with strong VSM practices. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.78

The greatest risk today isn't falling behind, it's pouring massive investment into chaotic activity that doesn't move the needle; VSM is the force multiplier that turns AI investment into a competitive advantage.

report: §Value stream management p73

reason: Both propositions appear near-verbatim in report#p74#0: "The greatest risk today isn't falling behind, it's pouring massive investment into chaotic activity that doesn't move the needle." and "value stream management (VSM) is the force multiplier that turns AI investment into a competitive advantage." The summary combines these two statements faithfully.

The greatest risk today isn't falling behind, it's pouring massive investment into chaotic activity that doesn't move the needle. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.74

value stream management (VSM) is the force multiplier that turns AI investment into a competitive advantage, ensuring that this powerful new technology solves the right problems instead of just creating more chaos. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.74

Between two people who share the same traits, environment, and processes, the person with higher AI adoption reports: higher individual effectiveness, higher software delivery instability, higher organizational performance, a higher percentage of time doing valuable work, higher code quality, higher product performance, higher software delivery throughput, and higher team performance.

so what: The report says Higher AI adoption links to gains across all listed outcomes, including delivery speed and stability. The source only says Higher AI adoption links to gains on most outcomes, but instability goes up, not down — a worse result.

report: §The results this year p38

reason: The passages confirm all positive outcomes listed, but software delivery instability is flagged as an undesirable increase (a negative outcome), not a neutral or positive one. The claim presents it alongside positive outcomes without that distinction, overstating the picture.

When comparing two people who share the same traits, environment, and processes, the person with higher AI adoption will, on average, report {number} more or less {outcome}. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.37

Higher levels of product performance • Higher levels of software delivery throughput • Higher levels of team performance — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.38

our data shows AI adoption not only fails to fix instability, it is currently associated with increasing instability. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.41

Note: An increase here is not a desirable outcome — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.38

it continues its detrimental relationship with software delivery stability. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.43

Higher AI adoption is associated with similar (no change in) levels of burnout and similar levels of friction.

report: §The results this year p38

reason: Multiple passages explicitly state AI has no relationship with burnout and no relationship with friction, directly supporting the claim that both remain unchanged with higher AI adoption.

It has no measurable relationship with burnout and friction, and it continues its detrimental relationship with software delivery stability. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.43

No relationship with friction • No relationship with burnout • AI is associated with an increase in software delivery instability — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.39

Similar levels of burnout • Similar levels of friction — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.38

Since 2024, AI's relationship with valuable time has reversed from negative to positive, its relationship with software delivery throughput has turned from negative to positive, and its relationship with product performance has shifted from neutral to positive.

so what: The report says AI went from hurting valuable time, throughput, and product performance to helping all three since 2024. The source only says Valuable time and throughput shifted from negative to positive; product performance change is not described.

report: §Changes in last year's patterns suggest adaptation p42

reason: Passages confirm the reversal on valuable work and throughput (report#p42#1, report#p80#0, report#p35#0). No passage supports a prior neutral-to-positive shift in product performance specifically linked to AI adoption since 2024.

This would explain why AI adoption starts to predict, in a reversal of last year's finding, a higher share of time in valuable work. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.42

AI adoption has now helped throughput tick upward, but instability still lingers. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.80

our 2024 DORA Report, which found that AI returned a lot of promising results but also increased software delivery instability and decreased software delivery throughput. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.35

AI adoption not only fails to fix instability, it is currently associated with increasing instability, and DORA found no evidence that AI adoption moderates the harms of instability on outcomes like product performance and burnout.

report: §Software delivery instability p41

reason: Multiple passages confirm AI adoption is associated with increased instability and that no moderating effect was found — instability continues to harm product performance and burnout regardless of AI adoption.

AI is associated with an increase in software delivery instability — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.39

We found no evidence of such a moderating effect. On the contrary, instability still has significant detrimental effects on crucial outcomes like product performance and burnout, which can ultimately negate any perceived gains in throughput. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.41

it continues its detrimental relationship with software delivery stability. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.43

Across different levels, AI is having a positive impact on most outcomes, with notable exceptions: it has no measurable relationship with burnout and friction, and it continues its detrimental relationship with software delivery stability.

report: §Conclusion p43

reason: Passages report positive associations with individual effectiveness, code quality, team/org performance, valuable work, and throughput; no relationship with burnout or friction; and a continuing increase in software delivery instability — matching the claim's subject, scope, and direction precisely.

Several outcomes continue to show patterns that suggest a less-than-favorable relationship with AI: • No relationship with friction • No relationship with burnout • AI is associated with an increase in software delivery instability — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.39

The effect of AI use on outcomes like throughput, code quality, and team and organizational performance was consistently amplified by seven team- and organization-level capabilities.

so what: The report says Seven capabilities always made AI's impact on every outcome bigger. The source only says Seven capabilities amplified AI's positive effects on certain outcomes, though effects varied by capability and some outcomes showed mixed or negative results.

report: §Looking beyond the tools to drive AI impact p80

reason: The seven capabilities amplified AI's benefits on specific outcomes, but not consistently across all outcomes listed. For example, quality internal platforms increased friction; small batches slightly reduced individual effectiveness; user-centric focus was required to avoid negative team performance. "Consistently amplified" overstates the uniform positive picture.

Of these, a set of seven AI capabilities showed substantial evidence of an interaction with AI use. That is, when teams paired these capabilities with AI adoption, the difference AI made across important outcomes was amplified. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.50

Conversely, we found that AI's neutral effect on respondents' reported experiences of friction is made harmful. That is, respondents experience more friction in organizations with quality internal platforms. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.62

Conversely, we also found that AI adoption's benefits for individual effectiveness are slightly reduced in teams that are working in small batches. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.58

Importantly, we also found that, in the absence of a user-centric focus, AI adoption has a negative impact on team performance. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.60

These systemic conditions reflect how an organization structures its work, supports its teams, and aligns its environment to modern development practices. These capabilities can help determine whether using AI tools translates into meaningful results, and that makes them amplifiers. The fact that they are all team- and organization-level reinforces a critical shift we need to make in how we think about AI's role in software delivery. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.80

AI's effects on performance depend on the system in which the work takes place; without foundational efforts to set up AI users for success, its benefits may stall, plateau, or stay unevenly distributed.

report: §The AI mirror p80

reason: Multiple passages confirm AI outcomes depend on organizational systems and capabilities, and that without the right environment benefits stay local. The claim's "stall, plateau, or stay unevenly distributed" is a reasonable paraphrase of these findings, though slightly more specific than the source's language.

More than anything, the environment that AI is nested in shapes its impact. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.80

In a system, improving one part does not guarantee better outcomes overall. In fact, local improvements can be blocked, diluted, or even reversed if the rest of the system isn't able to adapt. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.81

The overall pace of delivery is unlikely to change significantly unless the surrounding workflows are updated for developers' new tools and increased speed. The system isn't designed to carry the gains, let alone amplify them. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.81

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. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.3

We believe that the value of AI is not going to be unlocked by the technology itself, but by reimagining the system of work it inhabits. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.43

AI functions as both a mirror and a multiplier: in well-aligned organizations it amplifies flow, and in fragmented ones it exposes pain points.

report: §AI as mirror and multiplier p86

reason: Passages report#p87#1 and report#p87#2 state almost verbatim that AI is a mirror and multiplier, amplifying flow in well-aligned organizations and exposing pain points in fragmented ones. The claim faithfully reproduces the source's subject, scope, and direction.

This is why AI functions both as a mirror and a multiplier. It shines a light on what's working, accelerating what's already in motion, but it also surfaces what needs to change. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.87

In well-aligned organizations, AI amplifies flow. In fragmented ones, it exposes pain points. Teams that have strong practices, flexible workflows, and shared context often see immediate benefits. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.87

Deploying AI tools alone will not produce transformation; AI adoption needs to be treated as a transformation effort, with intentional changes to workflows, roles, governance, and cultural expectations.

report: §Organizations are systems, not sums of individuals p81

reason: Multiple passages explicitly state that deploying AI tools alone won't produce transformation, and that intentional changes to workflows, roles, governance, and culture are required — matching the claim's subject, scope, and direction precisely.

Deploying AI tools alone will not produce transformation, but paired with both pragmatic and visionary system-level changes, adopting AI can be a catalyst for reshaping how software is built, delivered, and secured. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.83

The earlier an organization begins treating AI adoption as a transformation effort, the more control it will have over how that transformation unfolds. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.84

As developers delegate more work to AI tools, tasks like verification, orchestration, and workflow design become more central. Organizations will need to define what those roles look like, how to support them, and how to align incentives accordingly. — report.pdf, Google Cloud / DORA, *State of AI-assisted Software Development* (the 2025 DORA, p.86