Time trackers come in two flavors. Both miss the point.

The category fails in two opposite ways, and Mark is built on a third option neither of them ships.

If you have ever tried a time tracker and quietly stopped using it, you have probably noticed something about the category. It argues about the wrong thing. Each camp positions itself against the other. Manual trackers like Harvest sell themselves on rigor, on the idea that good timesheets need a human in the loop. Passive trackers like Memtime and Timely sell themselves on relief, on the idea that you should never have to fill in another row again. They behave as if these are the two available answers to the same question.

They are answers to two different questions, and neither question is the one I needed answered.

The manual kind asks you to remember.

You open the app at the end of the day, or worse, the end of the week. You stare at a blank grid. You try to recall what you did on Tuesday between two thirty and four. You guess. You round up because you do not want to look idle, or you round down because you do not want to look bad at estimating. You attach a project code that feels close enough. You hit submit.

What you have produced is fiction. Not lies exactly. Just memory, run through the filter of how you want to be seen, rendered as decimals. The data is at best a sketch of what your week probably looked like if you squint. It is good enough for a client invoice because the client cannot see inside your head and does not actually want to. It is useless for anything else.

The product knows this. That is why the marketing copy never says "this will help you understand your work." It says "this will help you bill your clients." Those are extremely different promises. One requires the data to be true. The other only requires the data to be defensible.

The passive kind asks you to forget.

The pitch is exactly the inverse. Install a daemon. It watches your apps, your windows, your browser tabs, your meetings. It writes everything down. At the end of the week it hands you a perfect spreadsheet of where your hours went. You did nothing. You remembered nothing. The data is real.

The first week, you open the spreadsheet. You are mildly horrified by how much Slack. You make a note to do better. The second week, you open it less. By the third week, you are not opening it at all.

This is not because you are weak. This is because the data has no meaning attached. Nothing was meant by it. Nobody declared, at the start of the day, what the day was supposed to be about. So at the end of the day, the data has nothing to be compared against. It is just a record of what happened, in the same way that a CCTV camera in a corridor is a record of who walked past. You can watch the footage. There is no story.

The manual kind makes a story without data. The passive kind makes data without a story. Neither has both.

The third option neither of them ships.

The thing I wanted from a time tracker, after years of trying both kinds, was very simple. I wanted to declare, in the morning, what today was supposed to be about. Not a project code. Not a list of tasks. One sentence. The thing that, if I made it happen, today was a good day.

Then I wanted the tracker to quietly watch what actually happened. Not so I could be punished. Not so it could be sent to my boss. Not so I could feel guilt while looking at a pie chart. So that, at the end of the day, I could see the two stories next to each other. What I said the day was about, and what the day actually became.

The gap between those two is the only piece of information worth having. Everything else is bookkeeping.

I could not find a product that did this, so I built one. It is called Mark.

Why neither category ever shipped this.

Worth asking, because the answer reveals something about the category as a whole. Both kinds of tracker have had ten or fifteen years to add a morning intent ritual. It is not a hard feature. A prompt. A text field. A timestamp. Maybe forty hours of engineering and a designer's afternoon.

Manual trackers will not ship it because their business is timesheets. A morning intent declaration does not generate a billable hour. It generates a question, and the question makes the timesheet awkward. If the intent was "ship the migration" and the timesheet says "six hours in Slack," somebody has to explain the gap. Manual trackers cannot survive that conversation, because their entire value proposition is that the customer signs off on the timesheet without that conversation happening.

Passive trackers will not ship it because their pitch is "forget about it." The whole brand promise is that you never have to interact with the app. A morning intent ritual breaks that promise on day one. You have to interact. You have to think. You have to commit to something out loud. The passive tracker market is built on people who do not want to do any of those things. The feature is not missing because nobody thought of it. It is missing because shipping it would alienate the customer the product was sold to.

The third option has never existed because the two camps have spent fifteen years optimizing in opposite directions, and the thing I wanted is in the empty space between them.

What Mark does, specifically.

In the morning, when you open your laptop, Mark asks one question. What is today's mark? Five seconds. One sentence. You commit, or you skip the ritual and Mark falls back to honest tracking with no intent layer.

Throughout the day, a small native daemon watches what you actually work on. No timers to start. No tags to fill in. The data is real, in the passive-tracker sense, because nothing depends on your memory.

At the end of the day, Mark shows you the two stories side by side. What you said the day was about. What the day actually became. One number on the home screen, the focus ratio, expresses the relationship between them.

And because the data never leaves your laptop, a local language model can hold an unfiltered conversation with you about the gap. Why was Tuesday like that. What does the pattern of context switches say about how you are working. Whether the mark you set this morning is a real one or a thing you wrote down to feel productive. This is not a feature any cloud tracker can ship, because the moment the data leaves your machine, the privacy promise breaks and the architecture stops supporting the conversation.

The whole product follows from one design decision, which is that the data has to mean something before it gets captured. Everything else is the consequence.

What it is not.

Mark is not for billing, though it will give you the most accurate timesheet you have ever had. The data is captured continuously, classified automatically, and exportable. If your work is billable and you have spent years filling in Harvest from memory on Friday afternoons, Mark will quietly do better than that. The timesheet is a byproduct, not the point. The point is the gap between what you said the day was about and what the day became. If billing is the only reason you need a tracker, Mark works and is also overkill. Harvest is fine.

Mark is not for managers. There is no team dashboard. There will never be a team dashboard. Your manager will not open Mark and see your name. This is not a roadmap item. It is architecturally precluded by the local-first design, which is the point of the local-first design.

Mark is not for the device in your pocket. Apple already built Focus modes and Screen Time for phones. Mark watches your work. Apple watches your phone. Put the damn thing down.

If this is for you.

You will know. The feeling is recognition, not curiosity. You had stopped looking, and the description finally fits. You have probably tried at least one kind of tracker. You have probably stopped using it, for the reasons described above. You may have a Whoop on your wrist or an Oura on your finger because you already buy the premise that measurement, done honestly, changes behavior. Not because you collect dashboards. Because you wanted something to change. You may have read Cal Newport or Oliver Burkeman or Anne-Laure Le Cunff and felt something land that the productivity SaaS category never landed.

Mark is being built for you. It is private, native to macOS, single-user, and the data never leaves your laptop. There is no team plan. There will never be a team plan. You can sign up below for one email when it ships, and nothing else.