What this is
Work that originally took 5 days still takes 4.5 days after AI acceleration — individual output doubles, but team delivery cycles only shrink by 10%. This week's in-depth piece uses that number to argue the bottleneck in AI adoption has moved from "writing code" to "shipping code." The more compressed generation gets, the more prominent the costs at the handoff edges — review, alignment, status sync — get.
Industry view
In its 2025 annual report (DORA is the Google-led DevOps research program), DORA defines AI as an "amplifier" — pre-existing collaboration and architecture advantages get scaled up alongside it, which points the same way as the illustrative calculation above. The Team API concept from the Team Topologies authors (a software organizational design framework) is gaining more traction: package the team's interface docs, examples, status fields, and error handling into reusable "manuals," so consumers — and AI — can plug in self-service rather than pulling people into a group chat every time.
There are reservations, though. Some developers we track observe that faster submissions cause review queues to pile up, actually slowing overall cadence; others worry that once rules get stuffed into AI Skills (modules that bundle processes, operations, and tools for AI to invoke on repeat), stale content gets distributed faster. When docs, permissions, and validation mechanisms lag, even the smartest AI just relocates the wait from the generation side to the receiving side.
Impact on regular people
For enterprise IT: Rather than rolling out yet another AI assistant, inventorying a list of "who provides what, how to verify it, who maintains it" is often the better buy.
For individual careers: AI delivers clear personal productivity gains, but cross-team delivery pace may not change; when reporting contributions, "delivery completion rate" travels further than "output volume."
For consumer markets: The "how much AI sped up the company" rhetoric will thin out; "time to ship, time to verify impact" becomes the new evaluation anchor.