- Over half of engineering teams now consistently use AI coding tools
- Top adopters report double pull request throughput compared with low adopters
- Autonomous agents now handle an increasing share of routine coding tasks
The integration of AI tools in software engineering has shifted from experimental to operational, with more than half of engineering teams now relying on AI consistently, new research has claimed.
A report from Jellyfish claims nearly two-thirds (64%) of companies generate a majority of their code with AI assistance, showing a clear rise in adoption across the industry.
If current trends continue without interruption, this proportion could reach 90% within a single year.
AI adoption drives productivity gains
The incentive for this shift appears linked to measurable productivity gains rather than improvements in code quality.
“AI coding tools are now the default option for engineering teams, and the productivity gains are real,” said Nicholas Arcolano, Ph.D., head of research at Jellyfish.
This trajectory suggests that AI is no longer an auxiliary tool but rather the primary engine of software development for organizations that choose to adopt it aggressively.
While AI does not automatically improve code maintainability, the volume gains alone have made it the standard tool for many teams.
Top-performing companies in AI-driven sectors have experienced significant increases in output, and firms which adopt AI most aggressively report double the pull request throughput compared with low adopters over three months.
In practical terms, these teams are producing and shipping code at a pace that leaves competitors behind.
A rapidly growing trend within this adoption is the use of autonomous agents, which generate pull requests entirely without human intervention - although these agents currently make up a small portion of overall code production, their presence is expanding quickly.
In the 90th percentile of companies, contributions from autonomous agents rose from 10% of pull requests in January 2026 to 14% in February.
This indicates that AI-driven automation is not only supplementing human developers but is gradually taking on a larger share of routine coding tasks.
Despite these productivity gains, AI adoption does not guarantee fewer errors or improved code quality, therefore the focus for organizations has shifted toward monitoring operational output rather than assuming that faster production equates to better code.
For top engineering teams, AI’s value lies in its ability to accelerate development cycles and increase throughput.
As AI coding tools become the default in engineering workflows, top teams are completing tasks faster, and autonomous agents are taking on an increasing share of pull requests.
This shift affects how engineering teams plan, execute, and scale their work, and no team wants to be left behind for not following the trend.
For leaders, the focus is on integrating AI strategically to sustain high throughput, streamline operations, and maintain a competitive edge.
