For most of the last decade, business reporting worked one way. Someone needed an answer, so they requested a report. The data team built a dashboard. The dashboard answered that one question — until the question changed, at which point you got in line to request another one.

It's a strange arrangement when you say it out loud. We built a permanent, maintained software artifact to answer a question that was often a one-time curiosity. And we did it over and over.

The dashboard treadmill

Every contractor and platform we work with has some version of the dashboard treadmill. There's a backlog of report requests. There's a small number of people who can actually build them. And there's a growing pile of dashboards, most of which someone looked at twice and then forgot. Each new question spawns another report, and the backlog never shrinks.

We're now watching that loop break. On smaller datasets, we've started replacing Power BI reports with something far more direct — you ask the question in plain language and get the answer back, along with the logic it used to get there.

What changes when you can just ask

"Show me Q4 service revenue by business unit, and flag anything more than 10 percent off last year." That used to be a ticket. Now it can be a sentence. You get the answer, you can see how it was calculated, and you can ask the obvious follow-up without waiting in a queue.

Practitioners describe the mix flipping. Where reporting used to be something like 90 percent dashboards, some teams now describe it heading toward 20 percent dashboards and 80 percent questions asked in natural language. That's not a tweak. That's a different way of relating to your own data.

What dashboards are still best for

Dashboards don't vanish, and anyone telling you they do is overselling it. They shrink to the job they were always best at: monitoring a small set of core metrics that everyone watches all the time. Cash position. Backlog. Safety incidents. The handful of numbers that belong on a wall.

What goes away is the long tail — the hundred one-off dashboards built to answer a question that came up once in a board meeting. Those become conversations instead of construction projects.

The prerequisite nobody mentions

There's a catch, and it's the same catch that decides whether any of this AI stuff works: the answers are only as good as the data layer underneath. If "revenue" means three different things in three different systems, asking a question in plain language just gets you a confident wrong answer faster.

The companies that get this right won't be the ones with the most dashboards. They'll be the ones whose data is defined well enough that a plain-English question reliably returns a true answer. The reporting was never the point. The answers were.