Incredible First Artificial Intelligence Hackathon!

 

In Uruit we know we work with creative and talented people, but we’re still surprised when they come up with ideas like this one. Some time ago, we started working in a new initiative to build a tool that could help any team at Uruit add Artificial Intelligence capabilities to their projects.

That project became a new product called Cogniflow, a no-code Artificial Intelligence platform that makes it easier than ever for everyone to build solutions that involve Artificial Intelligence using text, image and audio.

With Cogniflow ready and waiting to be tested, what could be better than having our own teams use it to develop new Artificial Intelligence-based solutions ideas? Having them do it as part of a contest, with a jury that covered different areas, and a big prize!

We announced the activity and people got excited about it. It took only a few days for everyone to find a team and start working on their ideas. Finally, four teams signed up for the competition and the developers and designers that took part in the activity dedicated one full day to work on their projects; we called it “HackatonIA”.

Each one of the teams had to present their product to the members of the jury, paying special attention to the four areas that they evaluated: Platform mastery, Originality of the solution, Visual design, and Value proposition. Our special jury contemplated all the areas that we work with and consisted on one Product Owner, one UX/UI Designer, one Marketing Expert, and the three-people technical team that created Cogniflow.

Let’s take a look at the Artificial Intelligence Hackathon contestants!

WOLI: Written Offensive Language Identification

Internal chats are essential for any organization, and this is truer than ever now that so many companies have moved to remote work, and chats have become the main communication platform. This doesn’t come without risks, one of them being inappropriate or offensive messages exchanged between colleagues. Having that in mind, this team created a bot that identifies this kind of language and calls you out if you’re using it.

It’s simple, if you send something that catches the bot’s attention, you will receive a message that goes “I’m almost sure your message is offensive. Please stay calm and breath before you attack other people”. WOLI was trained in a week with around 1 million conversations, and it’s capable of identifying not only offensive words but also their context. The magic of Artificial Intelligence!

AAA: Android Application Analyzer

In this project, the members of the team used Artifical Intelligence to create a model that analyzes and understands app’s reviews in order to compute an implicit virtual rating. This was built in record time by people who had no experience in Machine Learning, thanks to Cogniflow the final product is a polished tool that can generate valuable conclusions.

CMN!: Cheat Me Now!

A developer and a designer partnered up here to create a fun platform with a little help from Artificial Intelligence. The game consists of a roulette with many sounds’ options, once the user runs it and gets a category, they have five seconds to imitate the sound and win. The players can get different kinds of sounds: animals, humans, domestic, urban or nature. When the players get it right, they get coins, and if they get it wrong, they can try again to succeed – and because it’s really fun! –.

The platform was trained to recognize all the sounds in the game: it not only tells you if your impression is accurate but can also say if it sounded like something else. For example, if you have to make a cat sound and you don’t nail it, the game might say something like “that sounded more like a siren”.

CDE: Car Damage Estimator

This was built by two developers, thinking about how Artificial Intelligence can help all the car owners that have no insurance. The final product was an app with many functionalities to make these car owners’ lives much simpler: it’s a place where you can estimate the cost of fixing your car, find qualified people to do that work, check information about your car to make informed decisions and much more!

The main attraction of the platform is the possibility to estimate the car’s damage; with only one picture, the app will say what is the affected area and if the damage is minor, moderate or severe.

And the winner is…

After the teams finished working on their projects and prepared a presentation, we gave the members of the jury some time to rate each creation. To make things more exciting and fun during these times of remote work and isolation, we organized an online event with a formal dress code, which of course only applied to the top half of our outfits, and once we were ready with our fancy clothes and slippers it was time to announce the winners.

First, we had a special mention for each of the teams:

• The WOLI project took home the recognition for mastering Cogniflow to achieve a great product.
• The prize for the most original solution was for AAA, because they found a unique way to solve a very specific issue and made it work amazingly.
• In the visual design area, CMN was outstanding with its polished design and UX.
• The CDE team won best value proposition, because the car damage estimation problem is an actual pain for many people, and they worked around that in an original and innovative way that impressed the jury.

And finally, it was time for the big prize! To maintain the gala spirit, we announced the winner of the Artificial Intelligence Hackathon with a video that was fit for the moment:

If you’d like to know more about Cogniflow, you can visit the website, and if you want to see the great digital products that our teams are able to build, you can visit our portfolio!