The 25 percent problem (McKinsey)

McKinsey's latest technology-trends session reports that only about 25 percent of companies using AI tools in their software engineering have significantly improved their pace or output.

That is a reported result, not a verdict on the tools. The session describes individual engineering work as fundamentally changed, and reported gains at the individual level vary widely. What the number says is that organisational improvement is the rarer outcome.

McKinsey's own diagnosis points away from the models: technology is often not the hard part, because change leadership, applied across governance, procurement and talent, is what decides whether a promising technology becomes a scaled business capability, and the operating model has to be redesigned with it, from skills and decision rights to accountability.

My own diagnostic, offered as a hypothesis rather than a finding, is to look at the queues. If engineers generate work faster inside an unchanged delivery system, the work arrives sooner at the same review, release and risk-approval steps, and acceleration pools in front of the slowest one. Where that is true, more acceleration at the code step buys very little. The place to start is measuring end-to-end throughput, from idea to production, before and after the tools arrive, and seeing where the time actually sits.

The conclusion follows from that measurement rather than from any benchmark. Before buying more acceleration for the step that is already fast, find out where the queue is, and who owns it.

#AI #Transformation #Leadership

Sources:
- McKinsey Live, "The technology shifts every business leader should be watching" (session recap, Michael Chui and Roger Roberts): https://lnkd.in/gSFyNg8s