AI Time Tracking: What It Is & Why We Don’t Do It

If you feel like AI is everywhere, you’re not wrong.

According to Forbes, 4 out of 5 companies prioritize AI in their business strategy. As consumers, we’re used to seeing products utilizing artificial intelligence. At this point, it seems like every piece of technology would benefit from AI-enrichment. Especially when it comes to tedious, overwhelmingly repetitive tasks like tracking time.

It takes 10-40 minutes to track project hours well if you do it manually, i.e., by setting up timers, coming up with task descriptions, and logging time into the system. The whole premise of Memtime is to eliminate these manual, time consuming tasks through automation. And yet we don’t call what we do AI time tracking (because automation does not equal AI).

In this blog post, we tackle the topic of AI in time tracking software at large and Memtime in particular.

What exactly is AI time tracking?

Let’s get 2 things straight.

1. Automatic time tracking ≠ AI time tracking

What we do at Memtime is called computer activity tracking. It’s a fully automated and completely private process of documenting one’s activity in all programs throughout the day.

There’s no artificial intelligence behind tracking computer activity. Memtime is a desktop app that follows your mouse and keyboard movement, therefore determining which program is running in the foreground. It then arranges your time and activity in all programs chronologically in a timeline.

For the purposes of explaining AI time tracking, it’s very important to emphasize that Memtime is a self-hosted aka offline time tracker. All activity data captured by Memtime is stored offline on the user’s device. We’ll come back to this point later.

2. AI time tracking = automatic time entries

Documenting how long it took you to work in certain documents and programs throughout the day is only one part of time tracking. The other part is determining which of those activities fall into the category of billable time and should be logged under tasks and projects in timesheets.

What’s referred to as “AI time tracking” is, in fact, automated project time entries. It’s only one part of the process that involves training the algorithm to guess which projects you’re working on. And then assign your time and activity in programs to those projects to save you time.

Spoiler: it doesn’t.

Examples of AI in time tracking software

Let’s now look at how AI is utilized in time tracking software available today. We’ll review 3 examples with distinct use cases, even though there are more tools out there that boast AI time tracking features.

1. Timely: automatic project time entries

Timely is similar to Memtime in that it uses a desktop app to automatically capture all computer activity. Except Timely’s Memory app doesn’t display the activity timeline, which can only be viewed via the web app, i.e., your activity data is uploaded to the cloud in real time as you work.

Timely’s AI then starts learning from how you log your captured activities under projects. The more you log, the better the algorithm gets at coming up with suggestions as to which activities should be logged under which of your projects.

2. Jibble: automatic timesheets + facial recognition

Jibble claims that it uses AI technology to record and categorize user activities across devices. This means that your activity data is continuously uploaded to the cloud and shared not only across your desktop and mobile devices, but also with the algorithm to train it to assign activities to projects and populate timesheets.

3. Timeular: calendar-based automatic time entries

Timeular is another automatic time tracker that captures computer activity and stores this data locally on your computer similar to Memtime. Like other AI time trackers, it automates timesheets by training the algorithm to guess which projects your captured activities belong to.

Why Memtime doesn’t use AI

Now that we’ve established what AI time tracking is, it’s time to clarify why Memtime doesn’t use artificial intelligence.

Privacy is incompatible with AI

To properly use AI in time entry suggestions, the algorithm needs to be trained in real time. As the user creates their project time entries, this data needs to be fed to artificial intelligence. The more projects the user has and the more unique they are, the more data is required for the algorithm to learn the patterns.

Memtime is a privacy-first automatic time tracker. We never upload user activity data to the cloud and never will. To achieve 80%+ accuracy in guessing time entries, we’d need to undermine our whole data safety concept, and even then the variety and scope of the user’s projects could be so vast the AI could still be wrong some of the time.

Not everybody wants AI in time tracking

It might come as a shock to some of the more tech-savvy readers of this article but not all people love the thought of artificial intelligence handling the data that directly impacts their billability and profitability.

In the early days of Memtime (formerly known as timeBro), we entertained the thought of adding AI to the mix, to save people even more time with automatic time entries. However, we soon found that most users aren’t comfortable with the idea of delegating billable time entries to an algorithm with compromised privacy.

AI creates more manual work

The core promise of AI-powered software is that it makes one’s life easier. The very reason people consider AI time tracking solutions is that they want the manual part of timekeeping to be handled by the algorithm. Simply put, they want to save time.

Counterintuitively, AI time tracking creates more manual work for the user. Consider this: if 80% of your activities are assigned to the correct projects by artificial intelligence, what happens to the remaining 20%? You need to go back and double-check your time entries to spot costly errors. From the first error you spot, you start investigating every AI time entry.

And then there’s the manual task of filling in the gaps. The 20% of inaccurately assigned time entries is never one consolidated timebox; they are scattered throughout the day. When you find an error, you also need to do the work of patching the time gaps around it.

Final thoughts on AI

There’s no denying that AI makes a lot of things easier, like coming up with meal ideas, writing code, or processing huge amounts of data in seconds.

This doesn’t mean that all software needs to be AI-enriched as some things require human touch, and some humans don’t make compromises when it comes to accuracy or privacy. Not to mention the fact that AI does not save time when it comes to project time entries.

We might revisit the topic of AI time tracking in the future. But for now, Memtime remains automated yet AI-free.