Last year, I wrote about Microsoft’s idea of the Infinite Workday.
Fast forward 12 months to 2026, that reality hasn't changed and AI usage at the workplace is far from being just experimental, it's in production at many corporates. Microsoft’s latest Work Trend Index 2026 goes further ahead and describes the rise of “Frontier Firms” reorganizing work around AI agents and automation giving us humans more “agency”.
We all know, the shift is real. Yet myself and most of you reading this have seen this happening before. Haven't we?
Agile began as a group of folks trying to describe a mindset to deliver better (and right) software. Later on, many organizations turned it into corporate shenanigans with a bunch of irrelevant velocity metrics, ceremonies, and extreme delivery pressure. We lost the chance to redesign the system and ended up accelerating the issues.
I'm generalizing here but what I notice nowadays are exactly the same patterns with AI adoption.
Instead of asking, “Should this work be done?”
We ask, “Can AI do it faster?”
If AI absorbs drafts, summaries, and can do a lot of the boilerplate work, what becomes of the apprenticeship layer? AI may erase the friction that builds expertise but at the same time destroying our capacity for thought and learning. Don't believe me? Read this recent discussion on reddit that showcases the issue.
AI is not the problem, Agile wasn't the problem.
Our operating model is.
Let me make it clear
Acceleration ≠ transformation. Speed at a broken system just scales dysfunction.
Efficiency creates expectation creep. Productivity gains often expand backlogs instead of reducing load.
Learning is at risk. With AI removing entry‑level work, we need to rethink how our capabilities are built.
What can you do (with an Agile mindset)
Remove before you automate / optimize. Commit to remove one recurring meeting or report or similar task for every AI workflow you introduce.
Redefine performance. Measure outcomes and impact, not responsiveness or output.
Rethink the apprenticeship model. Create learning paths that combine human mentoring with AI, not simply assuming capability will emerge automatically.
The technology might be new-ish. The behaviors aren't.
Moving forward, the litmus test won’t be whether you have AI capability.
It will be whether you are willing to remove work instead of just accelerating it.
