96.4% of surveyed organisations are using AI coding tools
GitKraken's 2026 State of AI in Engineering surveyed 554 developers and leaders. At that adoption level the interesting question stops being whether teams use these tools and becomes what it did to their process.
GitKraken's 2026 State of AI in Engineering report puts AI coding tool adoption at 96.4% of organisations, from a survey of 554 developers and leaders. It shipped alongside GitLens 19, which reworks the Commit Graph to put coding, review, coordination and shipping in one workbench.
At 96.4% the adoption question is closed. A number that high in a self-selected developer-tools survey should be read as "effectively everyone in this population", and the useful questions move downstream: what changed in review, in defect rates, in how long a junior takes to become useful.
The GitLens release is a decent proxy for where the pressure landed. Consolidating review and coordination into the same surface as coding is a response to volume — when generation gets cheap, the bottleneck moves to everything that happens after code exists. Review is now the scarce resource.
That is the part of this transition that is under-measured. Writing code was never the expensive part of software; understanding, reviewing and maintaining it was. A tool that accelerates the cheap step and floods the expensive one can raise output and lower throughput at the same time.
Which makes surveys like this most valuable for what they will show next year. Adoption is a lagging indicator of enthusiasm. Defect rates and review latency will be the leading indicators of whether it worked.
Agentic.ai — Agentic AI News — August 2026 Launches, Models & Research → · Morph — Best AI Coding Agents (August 2026): Scored Leaderboard →