AI coding assistants can produce code quickly and vendors promise large time savings, but most teams see underwhelming results and unchanged delivery timelines. Coding occupies about 16% of a developer's week and is the least friction-filled activity. The remaining 84% consists of information search, requirement clarification, meetings, and tech debt reduction, which create greater delays. Fragmented tooling forces frequent context switching across Jira, GitHub, Slack, email, and documentation hubs, causing mental resets and lost momentum. The most promising AI opportunity lies in automating and streamlining non-coding work around code to remove those hidden bottlenecks.
In the never-ending quest for developer productivity gains, a new default setting has been applied to engineering leadership teams: buy an AI coding tool. It's an understandable instinct. AI can now produce code in seconds, and vendors promise gains measured in hours saved per engineer, each week. But for most teams, the results are underwhelming. Delivery timelines barely budge, and the sense of "we're moving faster" fades.
The writing on the packet says AI coding assistants can save developers significant amounts of time. Coding only makes up about 16% of a developer's week, and it's the part they experience the least amount of friction. The rest of their work week is spent on more cumbersome tasks like searching for information, clarifying requirements, attending meetings, and finding time to pay down tech debt. By focusing on coding efficiency, we miss the glaring friction points that are slowing down the end-to-end software delivery cycle.
Collection
[
|
...
]