An analysis of 2,000 organizations worldwide finds that historically lower performing software engineering teams are seeing nearly a 50% improvement in lead time to delivery, a four times greater rate of improvement over higher performing teams that have adopted AI coding tools.
Conducted by Plandek, a provider of a software engineering intelligence (SEI) platform, the report also notes that despite that improvement, lower performing software engineering teams still deliver less than half the output per engineer compared to high-performing teams. For example, bottom-quartile teams still take more than 35 hours to merge pull requests, compared to under 21 hours for top performers, according to the report.
Top teams also ship software in under 22.5 days on average, while bottom-quartile teams take more than 62 days. High-performing teams complete over two-thirds of their planned sprint work, nearly twice as much as low-performing teams, which complete less than half and regularly miss the goals they set.
Higher performing teams also produce three times fewer bugs than the lowest performing teams and resolve bugs 45% quicker.
Additionally, high-performing teams spend more than 41% of their time on roadmap delivery, compared to less than 21% for low-performing teams whose capacity is consumed by bugs and unplanned work.
Plandek COO Will Lytle said that one reason AI is benefitting lower performing teams more is that there is clearly a number of best practices that AI makes simpler to follow. Higher performing teams, meanwhile, already encountered systematic constraints in their workflows that AI is not as likely to resolve without revisiting how their software development processes are constructed, he added.
The only way to identify those issues is to collect and analyze data from across the software development lifecycle (SDLC) that is needed to understand how best to optimize the underlying model relied on to build and deploy software, noted Lytle.
As low-performing teams continue to improve in the age of AI they will, of course, eventually encounter the same bottlenecks as higher performing teams. Regardless of software engineering maturity, AI will eventually alter the way software engineering teams are organized, said Lytle. Instead of being organized around small so-called “Pizza Box” teams, software engineering teams in the future are likely to consist of one or two software engineers augmented by AI agents that have been assigned to work on a specific product or project, he noted.
The degree to which that approach might reduce the overall size of the software engineering team will naturally vary from one organization to another. In some cases, organizations may decide to expand the overall size of their application portfolio, while others opt to reduce the total cost of maintaining their existing portfolio.
At the same time, the overall number of organizations that can afford to build and deploy custom software should increase as the pool of available software engineering talent increases.
It’s not clear how long it might take for software engineering workflows to evolve in the age of AI but the one thing that is certain is that a commitment to continuous improvement is now more crucial than ever.
from DevOps.com https://ift.tt/w6gnBAT
Comments
Post a Comment