AXON:OS is our proprietary software framework that allows us to rapidly develop applications that utilize artificial intelligence technology, in particular, swarm intelligence systems.

Based on AXON:OS, we develop various AxonAI products that are either stand alone, browser-based web applications, or integrated into application environments such as the Google playstore or the iOS app store. We also develop custom applications tailored to the requirements of specific customers through service contracts and licensing arrangements.

Broadly speaking, with our swarm intelligence components in AXON:OS we have the ability to reason over the relationships of large-scale sets of entities and to reason over causal chain of events that have happened in the recent past and make predictions of “what if” scenarios into the future. Out of these capabilities we assemble applications in three broad product families:



AXON:Investigate is a family of products that allow people in teams to explore large heterogeneous data sets, find information, and guide the assembly of models of the domain under investigation. Using machine-learning algorithms, our tools continuously estimate the intent of an investigation and, comparing it to available data, recommend relevant content to the investigator.

An example of such an investigation is, for instance, the investigation of patent landscapes in preparation of patent litigation or filing of patent applications. This is our very first prototype currently in beta stage. In this web-based application, the user queries a patent database, reviews the returned results, and judges which of the filed patents are relevant to the investigation. Those may then be bookmarked or expanded. Expanding a patent document enumerates a patent’s relationships to other documents through citations and through inventor relationships to researchers that have been working on that technical area. Thereby the investigator continues to expand the landscape, the graph of documents, people, and their relationships. The machine-learning component maintains a model of the investigation by analyzing the content of documents bookmarked or expanded by the user. That model is continuously matched against stored documents and strong matches are brought to the user’s attention.

The same task pattern – a continuous investigation of a complex topic against a large data set – can be applied to credit card transactions to investigate potential fraud, healthcare records to recommend treatment options or, again, to look at fraudulent activities, and many more such domains.


The second family of products in AXON:OS we call AXON:Anticipate. Here we have the ability to look at events that have happened in the past in a geographic environment and human terrain and apply model based reasoning to project what the risk of such events for the near future looks like. We have demonstrated that capability in prior work in IED risk forecasting. Our current application realized in AXON:OS forecasts the risk of violence events in Africa (e.g., Boko Haram attacks in Nigeria) with significant accuracy.


With the third family of AXON:OS products, AXON:Monitor, we continuously analyze live data streams from the real world. Those streams may originate, for instance, from a collection of devices that we want to monitor for abnormal behavior. Over time, we learn what is normal in the data streams and we apply statistical methods that identify deviations from this normal. With pattern recognition techniques, we may then classify those deviations as either benign or nefarious.

At the intersection of our forecasting capability (AXON:Anticipate) and the tracking of real-world data (AXON:Monitor) we create advanced cyber security tools that anticipate threats and actively defend against them.