Artificial intelligence
How developers and organizations are adopting AI tools and agents
The AI tool landscape
The tools developers are using or plan to try
AI is now the norm
1.01of employed developers use at least one AI tool at work
AI is now standard practice among developers, making AI adoption the norm rather than a differentiator.
The AI stack
1.02Adoption doesn’t mean coding professionals have settled on a standard stack; tool preferences vary notably by experience level.
Share of coding professionals who use each tool for work
Includes individual tools mentioned by at least 5% of respondents.
JetBrains Air Governance
Multiple AI tools. One place to manage them.
Developers use a range of AI tools. Air Governance brings access and spending management together, providing control in one place.
JetBrains Air GovernanceExperimentation with AI tools continues
1.03Widespread adoption hasn’t slowed experimentation. Developers are continuing to explore additional and alternative AI tools for their workflows.
Only 23% of employed developers don’t plan to try any new tools in the next year.
AI tools developers plan to try
Agentic development
How developers use and see AI agents in their workflows
AI agents are becoming part of the workflow
2.01For many developers, AI agents are supporting a significant share of their work.
More than half of employed developers are using AI agents in over 60% of their work tasks.
Share of development work tasks done with AI agents in the past month
Includes employed developers who use coding agents
JetBrains Air
Run any agent and review every change with JetBrains Air in IDEs.
Try it in your JetBrains IDEAgentic development practices
2.02Development practices when using AI agents, by years of experience
Includes employed developers who use coding agents
It’s all about context. Project guidelines (e.g. AGENTS.md, CLAUDE.md, .cursorrules) are the most common practice when using AI agents.
More experienced developers don’t just write prompts. They are more likely to give their agents guidelines, specs, and ways to verify their work.
The verdict on AI-generated code
2.03Less-experienced developers are more likely to think that AI agents produce better solutions than they would themselves. Senior developers are harder to impress.
Share of AI-generated solutions that are better than solutions written by …
Includes employed developers who use coding agents
60% of developers with 1–2 years of experience say AI-generated solutions are somewhat or much better than their own, compared to 47% of those with 16+ years. The difference by seniority largely disappears when comparing AI agents with other people (a colleague or “an average developer”).
Autonomous agents raise new concerns
2.04Why a company may not try autonomous AI coding agents in the next year, by organization size
Includes those who said “Not sure”, “Somewhat unlikely”, or “Very unlikely” about the company’s likelihood of trying agents that can autonomously complete tasks from start to finish.
Even as developers become comfortable using AI agents, organizations face additional barriers to adopting more autonomous agents. Half of developers say their company’s biggest concern is data privacy and security.
Adoption challenges vary by organization size. Smaller organizations tend to worry about the price tag, while larger ones face more governance and integration hurdles.
Shaping developer workflows
How the benefits and bottlenecks of AI are reshaping developers’ work.
Using AI
3.01AI tools are becoming integrated into core development workflows, especially around code generation.
How often developers are using AI at work for different tasks
Includes employed developers who regularly use at least one AI tool for coding and development at work
AI benefits
3.02Developers are crediting AI with helping them code faster, getting more repetitive tasks out of the way, and increasing their productivity.
The gains go beyond getting more work done. Nearly three-quarters of developers report increased productivity. 59% say AI helps them learn new technologies faster, and nearly half say it reduces the mental effort required for development work.
Benefits from using AI tools
Includes employed developers who regularly use at least one AI tool for coding and development
Shifting developer work
3.03The productivity gains from AI also bring workflow frictions. As AI takes on more work in generating code and documentation, developers’ work is evolving toward reviewing and correcting that output.
Share of developers spending more than a quarter of their coding time on:
Frictions reported from using AI tools in development tasks
Includes employed developers who regularly use at least one AI tool for coding and development at work
Drawing the line on delegation
3.04Share who are likely to delegate tasks to AI
Developers aren’t handing over the keys just yet. They are less likely to delegate tasks that require judgment and a deep understanding of the code, such as debugging or performing code reviews.
Adoption in organizations
How organizations are adopting AI and measuring its value.
AI in the workplace
4.01Two-thirds of developers say most coding professionals at their organizations are using AI developer tools.
The highest adoption level also shows the largest gap: 54% of developers at small organizations say more than 80% of their colleagues are using AI, compared to 37% of developers at large organizations.
Share of coding professionals who use AI, by organization size
Includes employed developers who work with other people
Encouraging AI adoption
4.02Organizations are taking a range of approaches to encourage AI adoption, but developers find the most value in practical support such as hands-on workshops, practical guidelines, and having a choice of tools.
Organization measures used to encourage AI adoption
Respondents could select up to three of their organization’s measures as the most useful
The AI payoff
4.03Productivity and speed are the most reported returns on investment from AI adoption, but those gains aren’t showing up equally across organizations.
Smaller organizations are less likely to measure ROI. At large organizations, tech managers are less likely to know what the payoff is.
Observed ROI from AI adoption, by organization size
Includes managers aware of their company’s developer productivity/experience practices
The measurement mismatch
4.04Metrics used to measure AI impact
Managers aware of their company’s developer productivity/experience practices, where the company tracks AI impact
What’s easy to count isn’t always what counts. Tech managers are more likely to see greater value in metrics that capture outcomes rather than usage.
Token usage is the most used metric for assessing AI impact, reported by 51% of tech managers. However, only 30% consider it among the most insightful. Net time gain has the reverse pattern: 39% consider it insightful, but only 28% say their organization tracks it.
ROI satisfaction
4.05Bigger organizations are still doing the math. More than half of tech managers at small organizations are satisfied with the ROI seen from AI spending, compared to just 34% at large organizations.
In some cases, ROI may be more feeling than formula: 40% of managers at organizations that don’t measure ROI still say they’re satisfied with it.
Satisfaction with ROI from AI spending, by organization size
Includes managers aware of their company’s developer productivity/experience practices
Fun facts
expect AGI within a decade. Only 14% say it will never happen.
When do you expect true AGI (human-like artificial general intelligence) will be developed?