Why I built the software my team runs on

I did not plan to write software. I planned to buy some.

The problem was ordinary. I had people working across five countries on different stores, and I needed to know who was working on what, how long it took, and what to invoice. Every tool I tried did roughly a third of that.

Time trackers tracked time and knew nothing about clients. Project tools knew about clients and could not produce an invoice. Screenshot monitoring existed as its own separate industry with its own separate subscription. And none of them agreed on what an hour meant, so reconciling three exports at the end of a month took longer than the work being reconciled.

The specific thing that broke it

Two tabs. A VA had the timer open in two browser tabs, and the activity tracking counted both. Not double the work, double the recorded minutes for the same minute.

It sounds trivial. It is not, because the number it produces is the number you bill a client. I could either explain to a client why their invoice was wrong, or I could stop using a tool that could not tell one minute from two.

So Tarly

Tarly is the thing I built instead. Time tracking, screenshots, activity, clients, tasks, invoices, a knowledge base. Ten people use it every day, which is the only reason I trust it: it gets used hard enough that when something is wrong, I hear about it the same afternoon.

The two-tab problem is solved in the least clever way possible. When the same person has the app open more than once, the tabs talk to each other and the visible time is taken as the maximum across them, not the sum. Two tabs open for the same minute is one minute of work. That is not sophisticated. It is just correct, which the tools I was paying for were not.

Would I recommend building your own?

Mostly no.

The honest accounting is that I spent evenings on this for months, and the subscriptions I avoided were cheaper than that time if you price my time at anything reasonable. It made sense for me because I wanted the thing to exist and because I can build it. If you cannot build it, buying four imperfect tools and living with the reconciliation is the rational choice.

What I would say is this: if a tool is producing numbers you have to correct before you can use them, that is not a small annoyance. That is the tool failing at its only job, and it will keep failing quietly until it costs you a client relationship.

We are building a second thing now, SWAPIT, for the automation side. It is not finished and it is not for sale. When it is real, it will be here.