The research was released on 30 September 2026. The pattern the data describes is not a straightforward dependency: 73% of respondents said they would be fine without AI tools. The qualification arrives immediately after, however, because that same majority had a blind spot. Forty percent had already pushed AI-written code to production that they did not fully understand, and 45% said they could not explain how code they had written themselves actually worked.
Yurii Nakonechnyi, co-founder and CTO at Sombra, placed this in a longer context: “That’s certainly an issue, but I’ve seen similar concerns back in 2010, when a lot of open-source application frameworks were created, and developers started using them without knowing how they worked under the hood. It made debugging or reviewing any code that depended on them hard. These days, the issue is that every codebase becomes its own self-built ‘framework’ built by AI, not a shared, documented one that everyone can learn.”
The analogy is imperfect. Open-source frameworks from that era were documented and community-reviewed, their behaviour known and discussable. AI-generated code that has not been fully reviewed before production creates a different kind of opacity: it is specific to one codebase, undocumented by default, and maintained by developers who by the survey’s own data may not fully understand it. Sombra works with software development teams across enterprise clients and conducted the survey across 2,000 developers who use AI tools in their daily work.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
