N-iX says AI speeds coding but not software releases

N-iX launched a new Engineering Index today and says AI coding tools cut build-stage time by up to 96%, while release cycles are still slowed by the same operational bottlenecks. The report, based on thousands of engineers across seven client programs, argues engineering leaders now need to fix measurement, security and decision-making gaps to see real gains.

Why it matters: - AI tools can make coding much faster, but N-iX says software delivery still gets stuck later in the process. - That means engineering leaders may see productivity gains on paper without faster releases in practice. - The report frames the real challenge in 2026 as scaling AI across teams, controls and workflows, not proving the tools can write code.

What happened: - N-iX launched the N-iX Engineering Index today as a new research series for engineering leaders. - The inaugural report, Pragmatic AI Software Engineering: A field guide for engineering leaders, is based on N-iX engineering engagements from Q1 2025 through Q2 2026. - The research covers thousands of engineers using AI coding workflows across seven structured client programmes. - N-iX says AI coding tools reduced time on piloted build tasks by 85% to 96%. - N-iX says software release cycles did not improve at the same pace.

The details: - N-iX says the main bottlenecks limiting release speed before AI were still limiting release speed after AI adoption. - Those bottlenecks include the measurement gap that can hide whether productivity gains are real, security review pressure from higher code volume, and slower human processes around requirements, acceptance and decisions. - The report says writing code used to be the slowest stage of software delivery, but AI changed that stage first. - N-iX says everything around engineering still runs at pre-AI speed, creating a new bottleneck. - A featured case study from a housing management technology company involved more than 150 engineers across 300 repositories. - In that case study, total annual incidents fell from 47,493 to 31,060, a 34% reduction. - Team velocity rose 94%, and the team now ships daily. - The case study also recorded a 34-point increase in test coverage and an 87% reduction in incident investigation time. - N-iX says adoption rate means the share of contracted developers showing at least daily AI activity on the contracted codebase. - The report also references independent sources on AI risk practices. - The full report is available here.

Between the lines: - The Index is aimed at organizations that already believe AI coding tools work and want help operationalizing them. - N-iX is positioning the report as a guide for board-auditable evidence, not just developer-level productivity claims. - The findings suggest the biggest ROI may come from fixing delivery systems around code, not from code generation alone. - The case study also shows that disciplined rollout, rather than isolated pilot success, is the harder problem.

What's next: - N-iX says the Engineering Index will keep publishing lessons from its client engagements. - The company says it will also surface the questions that remain open as AI adoption scales. - N-iX’s GenAI Value Lab says its APEX framework is designed to help engineering leaders solve the constraints that remain after coding becomes faster. - N-iX says the framework helped one client expand from 30 engineers across four teams to 100 engineers in three months.

The bottom line: - AI may have removed the coding bottleneck, but N-iX says release speed will not improve until companies fix the systems around it.