AIlon |
How it works: Understanding the world as a network

Perceiving the world in networks

How AIlon works

79% of the population use social media

Edison Research; Triton Digital

Most people use social media and their profiles actually reflect themselves and not an idealized version. This known, we started to develop AIlon, because most questions you might have about your audience, your clients, or the clients of your competition have already been answered - but in a latent form; in the form of digital footprints of consumers. We gather these datastreams every day and analyze trends and causalities.
Doing this, we have no competition as we are the only ones who work 100% plattform-independent, (thus we can make use of any social media plattform), work perfecetly compliant with the highest standards in the world regarding privacy rules (the European GDPR) and our Deep Learning algorithms are practice-proven, extremly precise and sensitive.


The vast majority of social media plattforms follow a grassroot-approach, which means that any user can create any "interest" or "page" to follow and any other user can follow it. Thus, it seems sensible to say that these pages and contents published, reflect anything that moves individuals.
The data we gather, are "following structures", which basically means: what set of pages does an arbitraty user follow on any social media plattform. Anykind of Personally Identifiable Information (PII) is not of interest for us, thus we completely anonymize any kind of information that could potentially lead to the individual behind the following-structure: This makes AIlon fully compliant with EU GDPR, which sets much higher standards than for example US privacy laws.

Image Simple illustration of an anonymized following structure

The illustration above visualizes how these following-structures look like. The aforementioned grass-root approach of social media plattforms explains why AIlon can analyze and predict any kind of "interest" that users might have.
As shown in the illustration, every kind of potential "follow" that AIlon gathered, is categorized into one of our more than 2000 categories, that are mostly a result of topic modeling algorithms, and they come with a website. This allows us to translate a plattform-dependent "follow" into a genric interest which is given by text. We apply Natural Language Processing (NLP) algorithms onto the texts and images of the website in order to make AIlon understand what actually stands behind a "follow". This makes following-structures first of all plattform-indepent, as the database turns into generic language and second, AIlon is able to investigate the semantic network of words, brands, preferences. This is also what makes the algorithms superior to others in the end: AIlon understands perceives as a network and learns from the entire network - not just your case.

Applying Deep Learning

As mentioned before, we apply Deep Learning algorithms onto these following-structures in oder to predict various attributes, based on the output of the NLP algorithms. For any follow-structure in our database, AIlon predicts attributes that fall into one of the following categories

  • Demographics
  • Psychographics
  • Geographics
Right now we predict approximately 120 distinct attributes.

Image AIlon can predicts attributes based on the anonymized following structure

We can predict basically anything you are interested in - also who is in your audience.

The point is, that the idea is generalizable: There is nearly nothing that AIlon can not predict.
What most companies are interested in, is how their clients and their audience looks like. Thus we retrain our algorithms in order to figure out which users are in your audience, which are your clients and which consumers buy at the competition and present these results in the dimensions of demographics, psychographics and geographcis to you.

Furthermore, AIlon's clustering algorithm can tell you how many segments he can find within your audience and clients and how they differ.

As we know what interests people in your audience have, we we also know how they will resonate to your activities.

Given the information which part of the population is your audience, we know what share of your audience follows which influencer, what newspaper they read or for example which sports club they support. Furthermore, we also know which kind of content they are more or less likely to react to. Drawing the big picture: we can deliver full algorithmic optimized marketing strategies.

Ready for a demo?

Contact us to get into augmented intelligence - better today than too late.

Please type your name.