Updated
Updated · O'Reilly Media · Oct 26
AI Reshapes Software Teams Into 5 Engineer Types as 40 PRs a Day Shift Bottlenecks
Updated
Updated · O'Reilly Media · Oct 26

AI Reshapes Software Teams Into 5 Engineer Types as 40 PRs a Day Shift Bottlenecks

3 articles · Updated · O'Reilly Media · Oct 26

Summary

  • Five recurring AI user modes are emerging on software teams—Soloist, Copy-Paster, Godot Waiter, Specialist and Tourist—as coding accelerates enough for one engineer to open 40 pull requests in a day.
  • That speed shifts the bottleneck from writing code to deciding why work should be done, with the report arguing roadmaps, issue framing and design reviews now matter more than ever.
  • The biggest risks differ by type: Soloists can flood teams with unaligned changes, Copy-Pasters can ship code they cannot explain, and Tourists can inflate adoption metrics without meaningful workflow use.
  • Historical evidence suggests speed alone can mislead: a 2025 METR study found 16 experienced developers took 19% longer with AI despite believing they were 20% faster, while a 2017 study found insecure copied code in 97.9% of sampled Stack Overflow snippets.
  • The report’s broader conclusion is that near-universal AI adoption in IT may reveal little by itself; the key divide is whether engineers still direct the tools or let the tools direct them.

Insights

If AI makes coding nearly instant, why are software teams suddenly drowning in slower reviews and hidden security risks?
Could the developer shipping the most AI-generated code actually be the biggest threat to your team's long-term architecture?
When AI agents act as non-human employees, who takes the blame when a machine accidentally leaks your company's sensitive secrets?