Bento

Bento is a machine learning software that allows businesses to accurately predict the cost of future endeavors and ways to optimize spending moving forward.

We have designed Bento to be very straightforward. The first step to using it is feeding it past data, such as costs of logistics, pricing of partners, and distributors; just to name a few. This will create a foundation for Bento to base its calculations on.

Now, consider this scenario: Your company would like to increase profits. To achieve this, you take a look at your spending. You might be tempted to switch to a different business partner or slim down your staff. But can you really know if it is a good idea?

The truth is, humans cannot. But Bento can. Through the user-friendly web interface, you can download all the necessary data for Bento to make an accurate forecast and give you favorable suggestions. The calculations usually take around an hour to complete.

Our programmers here at InventorSoft were tasked with creating this platform from scratch.

An early challenge we faced was that most machine learning libraries use extremely complex programs while not using an interface. This often leads to difficulties and misunderstandings.

In the end, we created a very flexible platform that could do any number of things. Additionally, it has a user-friendly interface. During the work process, it was very important to us to come up with an interface that was simple and could be used by anybody.

Creates a Bento account for the company. He can then configure individual user accounts and how much access they were granted.

Has access to specific machine learning models. He is also considered a user.

Accepts payment and oversees all registered users. Should there be any problem, he can step in and log on with any user’s account.

Can register on the platform and feed it with historical data.

We started off by building a solid backbone for Bento to run on. This included basic functionality and optimizing the registration process. Also, we added the graphics which visually represent what and how you can change for your benefit.

After that came the integration of loading models and training. For this, we used the Google Mail library. At this stage, we also set up the infrastructure to facilitate accurate forecasts. Also, we realized multiplier which gives the possibility to see how this or that case will behave at different values.

To round everything out, we added a payment system. We wanted to ensure the highest possible level of security and reliability. That is why we chose to incorporate technology from Stripe.