Atlassian report shows AI coding gains are offset by lifecycle friction for developers

Atlassian released a new report indicating that artificial intelligence is not solving productivity across the entire software development lifecycle. According to the findings, while AI accelerates code generation, it is shifting bottlenecks downstream rather than eliminating time constraints for engineering teams.

Nearly nine in ten developers surveyed stated that AI assists them in writing and shipping code more quickly. However, three in four respondents reported experiencing either the same level of friction or increased friction across the broader development process. This disconnect suggests that speed in one area is not compensating for delays in others.

The productivity paradox

Long-term optimism regarding AI’s ability to save time is declining significantly. The data shows that only 25% of developers currently save 10 or more hours per week using AI tools. This figure represents a sharp decrease from 68% in 2025. The report implies that initial efficiency gains are not sustaining or expanding into broader workflow improvements.

Atlassian Customer CTO Andrew Boyagi noted that faster code generation does not equate to better software delivered sooner. He explained that if teams continue to struggle with finding information, clarifying requirements, and moving work through the delivery process, speeding up coding alone will not translate into tangible business outcomes.

Strategic gaps and human bottlenecks

Human-centric tasks remain major obstacles in the development pipeline. Half of the developers surveyed identified finding information as a significant struggle, while another 50% cited meeting frequency as a primary pain point. These areas represent friction that current coding-focused AI tools do not address.

Many developers feel AI is being implemented too rapidly without adequate consideration for its real-world impacts. Three in five respondents said that merely using AI is viewed as a positive metric by their organizations, even when actual productivity gains are not realized. This misalignment between adoption metrics and operational reality highlights a strategic disconnect.

What should you do?

Organizations relying on AI for development efficiency should evaluate the entire software delivery lifecycle, not just coding speed. Leaders need to identify whether information retrieval and meeting overhead are negating the time saved by AI-assisted programming.

What to do and how to stay safe: Atlassian

• Audit your software development lifecycle to identify non-coding bottlenecks such as information retrieval and meeting frequency that may be offsetting AI gains in code generation.
• Review your organization's AI strategy to ensure it is clearly communicated and implemented, as only 17% of developers currently report having such a strategy in place.
• Measure actual time savings across the entire workflow rather than relying solely on adoption metrics, since faster code writing does not necessarily result in faster software delivery.
• Investigate tools or processes that address downstream friction points like requirement clarification and information access, rather than focusing exclusively on coding speed.

Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation

General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Is AI actually improving overall software development productivity according to Atlassian?

No, while 87% of developers say AI helps write code faster, 72% report the same or greater friction across the entire lifecycle, suggesting productivity gains are not holistic.

How has the amount of time developers save with AI changed recently?

The percentage of developers saving 10 or more hours per week using AI has dropped to 25%, down from 68% in 2025, indicating declining long-term time savings.

What are the main non-coding bottlenecks developers face?

Half of developers struggle with finding information, and another 50% cite meeting frequency as major bottlenecks that hinder overall workflow efficiency.