Last month Jellyfish released the industry’s most comprehensive quantitative analysis of AI transformation in software engineering. The results, compiled from the largest study of its kind, comprise real-world engineering signals about how hundreds of organizations and hundreds of thousands of developers are using AI tools across the SDLC.
This data set goes beyond simple activity signals to quantify the correlation between deep tool integration and its impact on measurable gains in delivery throughput and engineering outcomes.
We know that, based on our data – 700+ companies and 200,000 engineers over the last 13 weeks, November 2025 to January 2026 – pull request volume is increasing and cycle times are decreasing, but to understand AI’s actual impact on productivity, we need to look at features shipped.
To better answer this question, we divided companies in our study into four groups based on their level of AI adoption.
- Emerging: Companies where less than 25% of engineers are using AI tools regularly
- Developing: 25 to 50%
- Advanced: 50 to 75%, and
- AI-first: 75%+
We then looked at projects or epics and examined 1) AI adoption or the percent of frequent AI users in a company 2) epics shipped per 100 engineers and 3) median epic size to control for variations in project size.
Grouping companies by these cohorts and then looking at average epics shipped per 100 engineers in the last three months, we found that the volume of epics shipped increases with the level of AI adoption, up to a 56% boost comparing the top cohort to the bottom.
Effort per epic in terms of total person months also decreases as AI adoption increases, with an overall 25% reduction comparing the top cohort to the bottom.
Finally, the median number of tasks in each project was six across the board for all cohorts, showing that overall size of epics did not change, but shipping speed did increase.
Based on our data, it does appear that for companies who lean more heavily into AI use, they are in fact becoming more productive.
For more data and trends like this, you can check out Jellyfish AI Engineering Trends here.
About the author
Nicholas Arcolano, Ph.D. is Head of Research at Jellyfish where he leads Jellyfish Research, a multidisciplinary department that focuses on new product concepts, advanced ML and AI algorithms, and analytics and data science support across the company.
