Friday: An Intelligent Auto-Journal

Diary writing is an art of self-documentation that is virtually as old as writing itself. It is also a dying art. A respectable population of journal-keepers there may still be, but aside from these committed wordsmiths, society at large is simply not finding the time to keep a log. That’s a shame.

It also seems like an unnecessary chore. The huge quantity of digital data we produce on a daily basis, if collated, could provide a fairly accurate picture of our by-the-minute activities. Such an idea may seem somewhat futuristic, but this is the ambition which drove development company Dexetra to create the life-logging app, Friday. But this six-man team wanted to go beyond a simplistic journal; Binil Anthony, co-founder and CMO of Dexetra, outlined the vision to me — “to find info of an externality, we have Google and a lot of other search engines. But how do we search for things in our personal lives?”

A good question. With Friday, Dexetra intended to capture much of the data generated in daily phone use and combine it with the natural language processing engine found in their first app, Iris, in order to create a voice-searchable personal database. An incredible concept, for sure, but has it worked?

Like the article? You should subscribe and follow us on twitter.

Life Logging

So what exactly does Friday record, and how does it do it? The answer becomes fairly obvious as you view the list of permissions which accompany the app’s installation: phone calls, texts, voicemails, emails, photos, locations, music, social updates, startup and shutdown events, and more. Understandably, it takes a little while for Friday to compute all of the information and compile a back history based upon it, but once it does, it gives a shockingly detailed account of who you’ve interacted with, where you’ve been and what you’ve been doing. Remarkably, this is all effected with virtually no input from the user – at most, you may want to sign in to a couple of social networks. This is an app which truly lives up to its billing as an auto-journal.

Friday logs virtually everything you do.

Friday logs virtually everything you do.

If you’d like to further improve the accuracy of your Friday-based diary, you can also input your happenings manually via tweet-like, 140-character text updates, which you can optionally publish simultaneously on Twitter, Facebook, or Foursquare.

You can update manually — but will you?

You can update manually — but will you?

This, incidentally, raises an interesting question: why wouldn’t you just use your archived social output to keep track of what you’ve been up to? Anthony suggests that, “there is a real distinction between a social and real self.” But the likelihood that most users will bother to record their “social” and “real” thoughts separately, yet concurrently, seems remote.

That said, for those with sufficient personal enthusiasm for life-logging, Friday will certainly provide a pretty compelling historical record.

Browsing

By default, the data you collect via the various methods mentioned above is presented within the Digest section of the app. Here you’ll find a single, chronologically ordered column, allowing you to scroll back in time, and browse the content as you do so. You can also filter your history by data type, if you’re looking for something more specific.

The default method of display is the single column found in the Digest.

The default method of display is the single column found in the Digest.

Search

But the real engine room of Friday is its search function. “We tend to think that Friday is a tool to perform searches on personal life”, explains Anthony. The concept is that a user should be able to hunt down pretty specific information, such as “a picture taken at Central Park on Jan 20, 2011” or “what did I do in San Francisco?”

Natural Language Processing (NLP) has a big part to play here. Anthony tells me that he and his team wanted to make things easy for users of Friday, and “hence opted for a NLP (Natural Language Processing) based search over keyword search.” Dexetra implemented this technology by hacking Iris’ voice-search engine, which, “essentially utilizes the speech-to-text engine that is powered by Google.” The text is then passed on to a NLP engine, which understands the question and provides an accurate answer.

In use, this system works well. It doesn’t quite match the quality of recognition that Siri or Google Now can offer, but when it does understand, it brings up an accurate list of results fairly rapidly.

I can’t personally decide whether to embrace this all-knowing technology, or whether I find being confronted by a highly personal log of my life just a little too creepy. Each user will have their own interpretation, I suspect, but the technology is undeniably proficient.

Intelligence

The voice search is certainly clever, but there’s also some number-crunching intelligence here.

The app will, for instance, make suggestions, known as Insights, based on your previous activity. Friday will, for instance, prompt you to make a call to a contact “if it’s been a while since you got in touch” or, “if it makes sense to have a conversation in the current context,” according to Anthony. Friday also provides similar cues for music, places and weather.

Weather updates are one of Friday's more useful types of notifications.

Weather updates are one of Friday’s more useful types of notifications.

This, of course, takes us into the realm of computers trying to manage our lives. That isn’t a good space to be in, particularly with the relatively dumb computers we’re currently lumbered with. As a result, Friday’s prompts just feel weird, and their practical uses are limited.

Not that I think Anthony and his team mind this too much. All of their technologies have clearly been created with future opportunities in mind. Speaking about Natural Language Processing, for instance, Anthony notes that, “voice does look as if it will be the next level of user interface,” but, “the market is not quite ready for it now.” Equally, on commercial opportunities, he points out that, “anything that offers value would have a commercial angle that would eventually set in.”

It sounds to me like Dexetra is playing the long game.

Extras

There are, however, a couple of tricks up the sleeve of Friday which make the app pragmatically useful in its current form.

From within Friday, you can download and install “Applets” – Dexetra’s name for their range of apps which connect with Friday to share intelligence. Currently, there are two apps available; Dialapp replaces the default call history, providing “one-second dialling,” thanks to the same kind of predictions found in Friday, while Trails keeps a detailed log of your movements.

You can install 'Applets' to improve Friday's power.

You can install ‘Applets’ to improve Friday’s power.

Although they are not technically part of Friday’s feature-set, they add significantly to the overall experience.

Performance

I couldn’t sensibly feature an app that logs your every action without mentioning the battery juice it drains, especially given the excellent work Anthony’s team has done on this familiar technical challenge. He informs me that Friday would not figure on the battery list at all, despite its constant presence as a background app, and that battery usage remains somewhere between one and two per cent.

In real life usage, this means that there isn’t any notable drain.

Conclusion

Friday provides its users with a somewhat geeky diary, a searchable personal database, and some intelligence as to their phone-based communications. But, as with many things, these constituent parts don’t really describe the whole. Certainly, there are questions to be answered on practical use, although if you simply view Friday as an auto-journal, then you won’t be disappointed.

On the other hand, this is an app which feels like an experiment, or a platform for future development. Which isn’t to criticize it in any way as a current, standalone offering. But I just can’t help thinking that Friday is an app waiting for a mobile computing boost, or a major OS upgrade, for it to show its colours fully. When it does, it will be amazing –Star Trek-like, even.


Summary

A competent, voice-equipped automatic journal, with greater future ambitions.

7