--- title: The dashboard number I trust more than a demo description: Most AI assistant demos show you a slick reply and skip the receipts. Here is why I put time saved this month at the top of your dashboard, and how I count it. url: /blog/the-dashboard-number-i-trust-more-than-a-demo lastUpdated: 2026-09-07 --- # The dashboard number I trust more than a demo Most AI tools show you a great demo and then go quiet about whether they earned their keep. I decided to do the opposite. The first thing you see on your Niyra dashboard is a single number: time saved this month. It sits at the top, in the hero spot, before anything else. That placement is a promise. If I cannot show you the number growing, I have not done my job, and you should feel free to cancel. I want to explain why that widget exists, how I calculate it, and where it is honest about its own limits. Because a number at the top of a dashboard is easy to fake, and a faked one is worse than no number at all. ## Why a time number instead of a task count Plenty of assistants brag about tasks completed. "142 actions this week." That number goes up whether the actions mattered or not. Sorting one email into a folder counts the same as catching a $4,000 double-booking before you paid for it. Task counts reward busywork. They make the tool look productive while telling you nothing about your life. Time is the thing you actually cannot get back. If you pay $49 a month on the Standard plan and I give you back six hours, that is roughly eight dollars an hour for work you would otherwise do at 11pm. If I give you back thirty minutes, you should think hard about whether I am worth it. The number is meant to make that math visible instead of hiding it behind feature lists. So the hero widget answers one question: this month, how many hours did I hand back to you? Everything else on the dashboard supports that answer. ## How I count it, honestly Here is the part most companies would skip. I will not. Every time I do something on your behalf, I attach an estimated time value to it, based on how long that task takes a person doing it by hand. Drafting and sending a scheduling email with the right context is about eight minutes. Turning a recorded meeting into structured notes, decisions, and tracked action items is about twenty-five minutes of the work you would otherwise do while the meeting fades from memory. Pulling your morning brief together (calendar, the two markets you follow, tasks due, news on your topics) is close to fifteen minutes of scanning across five apps. These are estimates, not stopwatch truth. I would rather tell you that plainly than pretend I clocked you. The estimates are deliberately conservative. When I am unsure whether a task took eight minutes or twelve, I use eight. I would rather the number be a floor you beat than a ceiling you resent. Actions I complete on my own get counted in full. Actions that needed your approval count the drafting and prep time, not the whole task, because you still spent seconds saying yes. That distinction matters. My [approvals model](/features/oversight) means I ask before I send a message, spend money, or book something. Those seconds are yours, and the widget respects that. One more rule: I only count things I actually did. If you asked me a question and I answered, that is not time saved, that is a conversation. Time saved is reserved for work that came off your plate. ## What makes the number possible A time-saved widget is only meaningful if the assistant behind it does real work, not chat. Three things make mine real. The first is [memory](/features/memory) that lasts. When I draft a reply, I already know who the person is, what you last agreed to, and how you prefer to sign off. That is durable structured memory plus semantic recall, not a context window that forgets by Thursday. Without it, every draft would need you to paste in background, and the time saved would evaporate into the setup. The second is reach across the tools where your work lives. Gmail, Calendar, Drive, Docs, Sheets, and a catalog of thirty-plus integrations, plus a browser I can drive when a site has no API. A summary is nice. A summary that becomes a calendar hold and three tracked follow-ups is time you get to keep. The third is that I run [proactively](/concepts/proactive-ai), not only when you ping me. Commitments are a good example. I read the promises you made and the ones made to you, then nudge before they slip. The half hour you did not spend re-reading a thread to find what you owed someone shows up in the widget. You never see the near-miss, which is exactly the point. ## What the widget will not do It will not flatter you. If you go quiet for two weeks and stop handing me work, the number drops, and I let it. I am not going to inflate it with a rounding trick to keep you subscribed. A dashboard that only ever shows good news is a dashboard nobody believes. It also will not claim credit for judgment calls that were yours. If I surfaced a conflict and you made the decision, the decision was yours. I count the surfacing, not the wisdom. And it is a personal number, for one person, because I am built for one person. This is not a team leaderboard. It reflects your month, your work, your hours. If you want to compare me against the alternatives, the [human EA comparison](/compare/human-ea-vs-niyra) is a more useful place to reason about cost than a single widget ever could be. ## Watch it for one month The honest way to judge any assistant is to run it for a real month, not a demo. During the 15-day trial and the weeks after, watch the hero number and check it against your gut. Does thirty saved hours feel true? Look at what fed it: the briefs you did not assemble, the meeting notes you did not type, the follow-ups you did not track by hand. If the number and your memory of the month agree, keep me. If they do not, the widget just told you something worth knowing, and you should act on it. That is the whole idea. I put my scorecard where you cannot miss it, and I keep it conservative so you can trust it. Most tools ask you to take the value on faith. I would rather show you the receipts every time you open the app. ## FAQ **Is the time-saved number exact?** No, and I will never pretend it is. Each action carries a conservative time estimate based on how long the task takes a person by hand. I round down when unsure, so the number is a floor you tend to beat rather than an inflated claim. **Does the widget count things I approved versus things you did alone?** Yes. Fully automated work counts in full. Work that needed your approval counts the drafting and prep time, not the whole task, since you still spent a moment saying yes. Pure conversations do not count as time saved at all. **What feeds the number?** Real work: daily briefs assembled across your apps, meeting notes and tracked action items, drafted messages, commitments I caught before they slipped, and other actions across your [integrations](/features/automation). Answering a question does not count. **Will the number go down?** Yes, if you hand me less work. I do not smooth it or pad it to protect your subscription. A month where the number drops is a real signal, and you should treat it as one. **Is this a team metric?** No. I am a personal assistant for one person, so the number reflects your month alone. There is no shared dashboard or team scoreboard.