Engineering Role Shift: From Writing Code to Managing AI Output
Software platform Harness released its 'State of Engineering Excellence 2026' report on Wednesday, indicating that AI is taking over some coding responsibilities, shifting engineering roles toward managing AI output. The survey, covering 700 enterprise developers and engineering professionals, found that AI has become a default part of workflows, but teams face challenges in measuring productivity impact and ROI. Over half of respondents are concerned about performance evaluations based on AI data, hoping for improved separation between data and performance assessment clarity.
Quick Overview
- AI is reshaping engineering roles: technology takes over some coding work, and responsibilities shift toward managing AI outputs. The "State of Engineering Excellence 2026" report released by software platform Harness on Wednesday shows that the survey covered 700 developers and engineering professionals from enterprise companies.
- The report found that integrating AI into engineering workflows has become the default practice, but teams still face difficulties in measuring the impact of technology on productivity and return on investment (ROI). More than half of respondents said they are concerned about AI-based performance evaluations and hope for a clear boundary between data and performance assessments.
- Trevor Stuart, Senior Vice President and General Manager of Products at Harness, said AI is changing modern software development in ways no other technology has. "Engineering leaders are being asked to make multi-year AI investment decisions using dashboards built for a different era of software development," he said in a statement.
Deep Insights
AI is changing how engineering teams complete work and measure productivity, with developers spending more time on code review, bug fixing, and context switching between tools. When AI generates organizational code, output metrics improve, cycle times shorten, and developers feel more productive as work progresses faster.
But 81% of engineering leaders say the significant time saved on coding is now being spent reviewing AI's work. The report found that developers spend nearly a third of their day on this invisible work, which does not appear in productivity metrics like output.
"What organizations want to accelerate is not the work itself, but the overhead attached to the work," the report states.
Enterprises are also raising expectations for software developers. A report released last year by HackerRank found that more than two-thirds of developers said pressure to accelerate project delivery has increased.
Engineering responsibilities have expanded to include reviewing code quality and security, being accountable for downstream outcomes, and making judgments about when to trust AI and when to override its decisions. The Harness report notes that many enterprises have mature technology stacks to measure engineering outcomes, but no longer have the right tools to assess whether productivity gains are real.
The report suggests that technology leaders can take steps to update evaluation systems, such as tracking the organization's code delivery velocity or the time engineers spend reviewing AI outputs. Leaders can start by auditing the gap between what their organization's existing frameworks capture and the new demands created by AI adoption.
Similar to most AI adoption scenarios, Harness also recommends planning for more governance and security reviews; it also advises technology leaders to work with developers to establish systems and guardrails for measuring their own work.
Stuart said that until now, technological advances have not significantly impacted engineers and how they work. The rise of cloud and internet infrastructure operated beneath the developer role, but AI is forcing major changes.
"AI is completely reshaping the developer's job," he said, "and the measurement frameworks the industry has relied on for the past decade were not designed for this new unit of work."