News
AI is Speeding Up Coding, but Not Improving Engineering Outcomes
[UPDATED APRIL 27: Naomi Lurie, one of the authors of the report, added the following context in an e-mail]
"The report's central argument is that AI has become the primary author of code, but is not yet producing code at the level of quality that human engineers did — and that the engineering systems built around human-quality code were not designed to absorb what AI is now producing. The throughput gains are real. So is the quality deterioration. And the two are happening simultaneously."
[ORIGINAL STORY FOLLOWS]
Most enterprise engineering teams are now using AI tools in production, but a new report from Faros suggests many are not seeing corresponding gains in overall team performance because they have not adapted how software development is organized.
The Faros 2026 AI engineering report ("The Acceleration Whiplash") found that while AI is accelerating individual developer productivity, it does not automatically improve broader outcomes, such as product quality or delivery speed. The report highlights a gap between faster code generation and measurable gains at the team level.
“Software engineering with AI is a categorically different system from anything engineering organizations have run before,” said Neely Dunlap, a content strategist at Faros who focuses on AI and software engineering.
The report examines how AI is reshaping development workflows and asks what it means to operate software engineering systems built around AI. It concludes that many organizations treat AI as a standalone tool rather than rethinking processes throughout the full development lifecycle.
The report's authors allowed that developers are writing and shipping code faster with AI assistance, contributing to individual-level productivity gains. The tools in use range from integrated development environment assistants and chat-based systems to autonomous agents and specialized tools for code review and testing. And yet, they found that these gains do not consistently translate into improved team performance. Faster coding does not necessarily lead to better products or shorter delivery cycles, according to the findings.
They also said that many organizations simply lack clear visibility into how AI affects their operations, which creates blind spots in understanding where AI delivers value and where it may introduce new inefficiencies or risks. Instead of tech giants and startups flexing about "tokenmaxxing", which describes how much money they spend on AI tools like Claude and ChatGPT, they should be investing in both: skilled engineers and AI.
"To enable AI transformation, engineering leaders must have complete visibility into where AI usage sits across the organization, which licenses are dormant, and how AI is changing productivity, velocity, and quality across the business," the report's authors noted.
Key findings in the report show that:
The report suggests that organizations may need to invest not only in AI tools but also in engineering talent and workflow changes to realize broader benefits.
The AI coding landscape now spans IDE assistants, chat-based tools, autonomous agents, and review and test specialists. AI adoption is virtually ubiquitous, but companies are not accurately measuring it. The report’s authors argue that AI requires new ways of managing software engineering; there needs to be a shift around workflows, processes and performance tracking.
"AI is changing how engineering leaders build products and run their organizations, and the ones pulling ahead are synthesizing the pieces across all seven of these pillars rather than focusing on the two or three that feel most urgent this quarter," the report concluded.