One Year Later: What Actually Changed, and What Still Feels Unfinished

One Year Later: What We Actually Learned

This year does not start with a new promise. It starts with unfinished work. The question is no longer whether AI can be tested, but whether organisations can judge what those tests really showed. 2026 opens as a year of choices: what to close, what to fix, and what deserves the discipline of staying power rather than another experiment.

One year later, the main lesson is simple. Learning did not come from more pilots or smarter tools. It came from staying with what failed, asking why, and deciding what to stop. Where leaders treated pilots as work to be understood, learning accumulated. Where pilots were treated as signals of progress, learning stayed thin and quickly faded.

( read the article here.)