Original Research

The state of AI code.

We surveyed 105 engineering leaders about the AI code their teams ship, and what happens after it merges. Trust isn’t the only thing on the rise.
Download Report
Key Finding

Confidence in AI code is running ahead of verification.

78.1%

trust AI-generated code more than they did 12 months ago

61%

shipped a production incident originating in AI-generated code in the last 90 days

45.7%

saw AI-code issues drive up support load in the past 12 months
What's in the report

The data behind the headlines

Discover what's gone wrong, what it's costing teams like yours, and how to fix it.

  1. The trust paradox
  2. The failure record
  3. The review tax
  4. The adoption-verification gap
  5. The Context Void
  6. What this means