About Us Terms of Service Contact

News AI

Artificial intelligence algorithms separate interesting news from filler material.

Adding Intelligence to News

There are two problems we want to solve:

First, there problem of truly personalized news, a system which is capable of providing news filters for each user based on his or her interests, rather than some social network graph analysis.

The second problem we want to solve is to offer a mechanism for the financial compensation of the content providers you like to read and find interesting. This problem must be solved by means other than advertising. Not only we dislike ads, but we also know that he who pays the piper calls the tune.

You will find that many of our sources are distinctly not mainstream. If you are satisfied with mainstream media, you know which newspapers to subscribe to and what channel to tune to. News AI is intended for people who need to know more, who want to decide for themselves, or simply who do not like to have their choices limited by advertisers, censors, or editorial policies.

We solve these two problems by offering a service which uses artificial intelligence (AI) algorithms to read, analyze, and rate news articles according to your preferences. When the beta period is over, we'll ask you to purchase a subscription, the large fraction of which (perhaps 80%) will be sent to the content providers in proportion to the number of articles you read. This way, we believe, we can feed two birds with one scone provide this service for you and support the authors at the same time. We do not serve advertising, we do not re-sell your personal data, and we do redirect to the original article URL to preserve the authors' advertizing revenue.

Quick Start

First, download the java application (you may need to install Java). After registration (email us for an invitation during the beta testing period) you will see your news stream. The AI assigns a rating number to each article according to how much it thinks you will like (+1.0) it or find it not interesting (-1.0).

Double click on the article you find interesting. This will open the original article in the browser, and also signal the AI that you found something in this article interesting.

Conversely, click on the "x" icon on the left side of the article to signal the AI that you are not interested.


When the user rates an article, the algorithm tries to determine which patterns triggered the user's decision. The patterns come from the article itself as well as the user profile being built by the algorithm during the interaction with the user. With time, the algorithm is able to rate article with more and more accuracy. More interesting for articles tend to get higher rating, the less interesting ones go into negative territory.

There is nothing "social" in the algorithm. The recommendations generated by the system are not some average of some user group, the results are generated specifically for each user.

Known Issues

Presently, duplicate articles are not detected and filtered (apparently, copying and re-transmitting news is a popular past time).

The current prototype application requires java and as a result is not available on tablets or smartphones. An iOS or Android app will be much more convenient.

Download Reader Application

Download NewsReaderApp.jar for OS X, Windows, Linux.

Android/iOS App

We are considering releasing an Android and an iOS app.