posted 10th September 2026
AI has automated the build and test work that agile coaches used to sit in the middle of. Scott Howard has spent two years working out what is left, and it is not what most of the community expects.
Scott is a former colleague of mine from NatWest Group and a technology consultant since the dot-com era. He started in the communications side of tech change in the late 90s, built a web 1.0 department off the back of a Bill Gates speech, and has spent 25 years translating between business people who cannot articulate what they need and engineers who just want to be told what to build. For the last couple of years he has gone deep on AI, including hands-on work with a machine learning team at a renewable energy operator.
His headline: AI is the biggest gift to our community in a long time. Machine learning starts from failure and improves through fast iteration, which is the mindset agile coaches have been trying to sell to leadership for years. The catch is that in regulated industries, "celebrate failure" gets you a dirty look, because if you fail, someone loses their job.
We get into what that means for the job. The value shifts hard left into use-case backlogs, data quality and risk, and hard right into testing and outcomes. The old middle ground of coaching teams through build and test is gone. Put rubbish into a system that fast and you do not get a trickle back, you get a flood, out of a black box you did not build and cannot debug.