Recommended Free Tools
The strongest AI-assisted engineering teams do more than give developers a coding assistant. McKinsey’s 2025 survey found that its top-performing fifth were more likely to use AI across the software development lifecycle, reshape roles and release ownership, train people hands-on, track business and engineering outcomes, and actively manage organizational change. These are patterns associated with higher performance—not proof that any single practice causes it.
What does “top 20%” mean in this comparison?
In its November 3, 2025 analysis, McKinsey surveyed nearly 300 senior leaders at publicly traded companies. One hundred assessed AI’s performance impact across software quality, time to market, team productivity, and customer experience. McKinsey defined “top performers” as the top quintile across those four measures and “bottom performers” as the bottom quintile. Respondents represented the Americas, Asia, and Europe, as well as several sectors. McKinsey’s analysis and methodology therefore describe reported organizational outcomes, not a controlled experiment or a universal ranking of engineering teams.
| Reported comparison | McKinsey’s finding |
|---|---|
| Difference between top and bottom groups | 15 percentage points in performance across the four outcome measures |
| Top performers’ reported improvements | 16–30% in team productivity, customer experience, and time to market; 31–45% in software quality |
These ranges are what top performers reported; they are not forecasts for a new adopter. The analysis does not establish that every company measured outcomes identically or that the practices below caused the performance gap.
What higher-performing teams do differently
They apply AI across the lifecycle, not only to coding
McKinsey found that top performers were six to seven times more likely than peers to scale at least four AI use cases across software development. Nearly two-thirds of leaders reported four or more use cases at scale, compared with 10% of bottom performers. The use cases span design, coding, testing, deployment, and adoption tracking. The point is breadth across the workflow, not maximizing the number of tools or forcing AI into every task.
Free tools Windows power users keep installed
One-click scans. No signup required.
#1 Best Overall
A practical starting point is a real bottleneck—such as slow test authoring or a hard-to-navigate codebase—then an evaluation of whether AI helps there. Expand to other lifecycle stages only when the team can verify the result and manage the risks.
They evolve roles and make delivery ownership clear
The higher-performing organizations McKinsey describes are changing how work is divided. Engineers may take broader responsibility for product decisions, architecture, testing, and AI-assisted workflows, while release ownership becomes clearer. This matters because faster code generation alone does not decide whether a change is safe, useful, or ready to ship.
McKinsey’s discussion of Cursor is an interview-based illustration, not a controlled comparison. The broader lesson is to define who reviews AI-assisted work, who makes product and architecture trade-offs, and who owns the release rather than assuming a tool will settle those responsibilities.
Rank #2
They teach through work, not just tool demonstrations
Hands-on workshops and one-to-one coaching were reported by 57% of top performers, versus 20% of bottom performers. McKinsey recommends grounding practice in real engineering activities such as code review, sprint planning, and testing. That kind of training lets people learn where AI is useful and where they need to verify or reject its output.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Make time for practice within normal work, and give teams examples tied to their own systems and standards. A one-off tool introduction does not substitute for learning how to use AI responsibly in the workflows the team actually owns.
They measure outcomes instead of treating activity as value
McKinsey reports that 79% of top performers tracked quality improvement and 57% tracked speed gains. Those outcome measures are more informative than code volume or the share of code generated with AI, which can rise without improving software or delivery. The analysis recommends pairing outcome measures with input measures such as tool adoption.
Rank #3
- Delivery: cycle time and time to market.
- Quality: release quality and software defects or other quality indicators the organization already tracks.
- Customer impact: customer experience or satisfaction.
- Adoption: whether and where teams use the tools, interpreted alongside the outcomes above.
Set a team baseline before changing workflows, then compare later results using the same definitions. This does not by itself prove AI caused a change, but it makes a local claim more meaningful than a usage count alone.
They manage change and reward useful behavior
Nearly eight in ten top performers connected generative-AI goals to both developer and product-manager reviews. By comparison, 10% of bottom performers did so for developers, and none did so for product managers. McKinsey advises rewarding contributions such as identifying worthwhile automation opportunities and improving quality, rather than raw tool usage. This keeps incentives closer to useful results than to activity that is easy to count.
What organizations should put in place
DORA’s January 2025 guidance, updated March 19, 2025, recommends making AI adoption understandable and workable for developers: communicate the organization’s plans, address concerns, make time to learn, and establish usage policies. DORA’s analysis included 1,000 developer and developer-adjacent respondents and used Bayesian regression on self-reported team AI usage. Its findings concern associations with adoption; they should not be read as universal causal effects or productivity gains.
Rank #4
- Explain the plan. Tell teams what the organization is trying to improve and how it expects AI to fit into existing work.
- Surface and address concerns. Give developers a way to raise practical questions about the workflows and policies they are being asked to use.
- Protect learning time. Let people practice on real tasks rather than expecting adoption to happen alongside unchanged workloads.
- Set clear usage policies. State what uses are allowed and what developers must check before relying on AI-assisted output.
- Assign ownership and evaluate results. Establish responsibility for releases and workflow changes, then track delivery, quality, and customer outcomes alongside adoption.
Why high tool adoption is not the same as high performance
AI coding tools have become common in some large-company settings, but prevalence does not show that teams are getting better results. GitHub’s online survey included 2,000 non-student respondents at companies with at least 1,000 employees: 500 each in the United States, Brazil, Germany, and India. Conducted February 26 to March 18, 2024, and updated April 15, 2025, it found that more than 97% had used AI coding tools at work at some point. The survey did not measure how frequently they used them, and its self-reported findings cover those four markets—not all engineering organizations. GitHub’s survey details make it useful as adoption context, not evidence of team-level causal gains.
DORA’s 2024 report drew on responses from more than 39,000 technology professionals worldwide and discusses AI alongside platform engineering, user-centricity, and stable priorities. That broad report supplies organizational context; its abstract does not substantiate a specific claim about which individual AI practices produce better results.
How to apply the findings to your team
- Choose a development bottleneck that can be observed, and start with an AI use case aimed at it.
- Give the team practice time, policies, and clarity about ownership before expecting a new workflow to stick.
- Measure a baseline and follow-up for delivery, quality, productivity, or customer outcomes using consistent definitions.
- Expand to other lifecycle stages when a use case proves useful and its output can be checked.
- Describe local improvements as observed results, not as guaranteed effects of AI or proof that one practice caused the change.
The distinction is organizational: the higher-performing teams in McKinsey’s comparison combine lifecycle-wide use with changes to skills, roles, measurement, and incentives. The percentages identify patterns worth testing in a team’s own setting—not a recipe that guarantees the same gains.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
