Why we deleted our own dashboards.
A dashboard is a question you have to remember to ask. We replaced ours with the surface our AI employees already work on: a queue of items that arrive with the evidence attached.
We deleted our internal dashboards. The funnel board, the usage grid, the pipeline view someone built in a weekend and we all bookmarked. Gone. We replaced the wall of charts with the same surface our product is built on, a queue of items that come to you with evidence attached, and we got faster, not slower.
A dashboard is a question you have to remember to ask
A dashboard does nothing until you go look at it. It sits there, perfectly up to date, waiting for you to remember it exists, open it, scan it, decide which of the twelve numbers moved in a way that matters, and then work out why. Three jobs, and the dashboard does none of them. The failure mode is not that the chart is wrong. The failure mode is Tuesday afternoon, when the thing that mattered started slipping and nobody had the board open, because nobody has a board open at the moment something starts to slip. You find out at Friday's review, by which point it is not news, it is a postmortem.
Q: which accounts churn this month?
A: Globex and Northwind. Save plays are drafted and waiting on you.
A queue is work that comes to you
The replacement is not a smarter chart. It is a different shape entirely. Changes that need a decision arrive as items in a queue: what moved, why it matters, and the sources under each claim, ranked so the top of the queue is the thing to look at first. This is the same shape our AI employees work in. They watch their own inputs, work what arrives, and escalate to a person as a queue item with a journal, never as a chart someone might eventually read. We run our own company on that surface because we are asking customers to run real work on it.
By Friday's review it is not news, it is a postmortem. The point of a queue is to catch it on Tuesday.
The source is what makes push survivable
Here is the trap with any system that tells you what matters: the first time it tells you something wrong, you stop believing all of it, and now you are back to checking the dashboard, except you also have an inbox of pings you ignore. A push you cannot verify is worse than a pull you can. So every line of every item is clickable down to where it came from. The number opens to the query. The claim opens to the record or the page. A system that summarizes and asks for trust is a liability. A system that summarizes and shows its work is a tool.
“Globex is the renewal most likely to churn this month.”
sourcesWhy this stays approval-first
Flipping from pull to push does not mean handing the wheel to the software. The system notices, gathers, and proposes. The consequential call waits for a person, every time. Push the information freely. Never push the irreversible action without someone saying go.
We are not anti-data. The sources sit under every claim. We are anti-homework. Make the system come to you, make it show its work, and make it wait for your call on anything you cannot undo. Then delete the dashboard. We did, and we are not putting it back.
Hire an AI employee for one role, watch it work a visible queue, and approve every output before it counts.
