New: the shifts that come back across the latest research. See what to focus on

Horizon: now

Software teams shift from coding to directing AI

Software teams increasingly direct AI instead of writing code, making shared knowledge, clear responsibilities, and efficient checks essential for reliable delivery.

In the report's words: “Software teams move from writing code to directing AI agents, requiring better context, checks, and clear human ownership (p3, p12).”

The number that matters

27%p5

Organizations with tools to fully use generative AI

Where to focus

  1. 1Train teams and clarify responsibilities for AI codingp5
  2. 2Write down design decisions and business knowledge for AIp11
  3. 3Speed up reviews of AI codep9

The details

Who it affects

Engineering leads and senior developers at large enterprisesp12, p20

What holds them back

  • Engineering leads struggle to scale AI coding because teams lack the skills, role clarity, and supporting toolsp5
  • Developers struggle to get reliable AI code because design reasoning and business knowledge remain in people’s headsp11
  • Engineering leads struggle to deliver faster because manual reviews of AI code become bottlenecksp9

The numbers

  • Only 27% of organizations have the platforms and tools needed to fully use generative AIp5
  • Every 25% increase in AI adoption correlates with a 7.2% rise in delivery instabilityp14

What people search for

  • how to roll out AI coding tools to my team
  • AI coding training for developers
  • how to give AI context about my codebase
  • how to document code design decisions
  • automate AI code review
  • how to test AI generated code