About us

Semantic Visions is a software-based actionable analytics firm based in Prague and London, operating a military-grade Open-Source Intelligence (OSINT) system that collects and analyzes 90% of the world’s news content. Founded in 2011, Semantic Visions runs a cross-language early warning system to protect clients from threats before they materialize. Semantic Visions is committed to safe-guarding democracy through the detection of disinformation and adversarial propaganda, and by fostering joint situational awareness of emerging events and trends.

Read more

Our Story

In 2005, when Frantisek “Frank” Vrabel was working in the defense industry on C4i (Command, Control, Computers, Communications, and Intelligence) systems technology, he recognized an opportunity for expanding the use of Open-Source Intelligence (OSINT) collection and analysis. At the time, militaries and intelligence agencies – historically the institutions leading the way in digital innovation – were not using cross-language semantic analysis in their OSINT systems, which created huge blind spots for operational readiness. Vrabel knew there was a better way to put public information to work. So, in partnership with a few trusted data scientists and software developers, he set out to provide it.

Part of the motivation was personal. Vrabel grew up in Iraq, where his father was a visiting professor at Baghdad University, and was acutely aware of the hazards in poorly designed OSINT collection methods. In his estimation, most of the misjudgments made by the US military in the run-up to the Iraq War in 2003 were due to a narrow understanding of the local context based on an over-reliance on English content. Case in point: in one Iraqi city, the US military built a base atop ancient Babylon, a cultural misfire that better Arabic contextual analysis could have prevented.

A young Frantisek "Frank" Vrabel, Founder of Semantic Visions, seated atop the Lion of Babylon, 1968

About us

How it works

The opportunity

To improve OSINT’s reach and deliver threat-detection capabilities to the civilian sector, one of Semantic Visions’ first objectives was bridging the so-called English language divide.

While English represents about half of all internet content (and English-speaking economies produce roughly 25% of the world’s GDP), only about 10% of global actionable intelligence exists in English.

That’s because most events surface locally, in languages other than English. Thus, the capability to monitor threats in multiple languages and establish full operational visibility is essential for organizations with a global footprint.

The solution

To address the first challenge, we began compiling a database of online media, webpages, and blogs. With our database packed with millions of sources and billions of articles, we then sought to mimic in code how humans derive meaning from language – through sentences, not keywords.

We then built a universal semantic language by creating computer-processable identifiers that have the same context irrespective of language. With this common semantic language in place, giving us the ability to monitor threats in English, Chinese, Russian, Arabic, Korean, Spanish, German, French, Italian, Portuguese, Czech, and Japanese, we then wrote code to search for and compare more than 500 risk types, a catalogue of threats ranging from Child Labor to Environmental Sustainability.

To overcome the significant technical challenge of distinguishing signal from noise, we used our proprietary AI solution based on advanced semantic analysis (ASA) and big data semantics (BDS). ASA generates granular and precise metadata based on this cross-language principle, while BDS synthesizes scenarios from entities (geolocation, organization, company, and person), events, sentiment, and context. These scenarios can encompass multiple events and are often triggered by signals coming from multiple articles and languages.

Today, our database is one of the largest and most comprehensive collections of online news anywhere, and the events we are able to detect are constantly evolving. Few risks slip through the cracks, even if they are previously unknown. For example, while we didn’t have the terms “COVID-19” or “coronavirus” in our risk maps at the time of the outbreak in early 2020, our software did detect an emerging trend related to “SARS”-like virus, “infectious disease,” and “respiratory disease” in late 2019, well before the pandemic reached Europe.

Fifteen years after its conception, Semantic Visions continues to illuminate evolving risks and threat-related trends. With Semantic Visions, customers have a powerful ally in the forever war on risk.
Let us help you with your security and enterprise needs.
Get in Touch

More than

280

Companies trust
our solutions

7+

MIL

Entities we monitor
in real time

Get in Touch