Рубрики ArticlesInterview

Turns chaotic data into goals for the Armed Forces. How Kyiv activists created the Griselda analytical system

Опубликовал
Алена Лебедева

Activist and public figure Dmytro Shamrai together with like-minded people, created a system Griselda. It uses artificial intelligence to collect, process, and transmit huge amounts of information to the military in the form of intelligence. In 2022, the system processed almost 250 thousand different pieces of information, and in 2023, the military used it to receive about 4 thousand targets every week.

The MC.today editorial team spoke with Dmytro. In the interview, we talk about how he and his friends turned from ambassadors of medical cannabis to founders of a system that has been helping Ukrainian defenders destroy the enemy for three years.

Before the full-scale war, he fought for medical cannabis

Before the full-scale war, I was a public figure on «full-time» For 15 years, I have been involved in the regulation of medical cannabis and the humanization of our Ukrainian drug policy in line with global and European values. I still believe that the issue of access to effective medicines is only becoming more urgent.

I am one of the board members of the organization Hemp Freedom MarchHe is also a member and active participant of the «Democratic Axe» party, a member of the public council at the State Service of Ukraine on Medicines and Drugs Control.

In 2015, like-minded people and I launched the project «Benefactor». It is an all-Ukrainian online platform where users can post a request for help, respond to it, or offer their assistance to others free of charge In fact, it is a social network for mutual assistance.

I have always been concerned with socially important things. For example, I believe that we are responsible for our city and society. For the space we create and where we have to live. That’s why I spent a lot of time building civil society institutions. I did this until February 24, 2022, and I continue to do so.

There has been no doubt that there will be a war for a long time

In the winter of 2022, there was a general feeling in the air that the stakes were too high for Russians. Society in Russia was so wound up that it was necessary to channel these emotions somewhere. If they weren’t channeled outward, they would turn inward and turn against their government.

So a couple of weeks before the full-scale war, I started looking for a place where I could be useful. At the end of February, my friends and I became members of the Kyiv DFTG (Volunteer Formation of the Territorial Community).

Our headquarters was in the Kyiv City State Administration, and we were doing a lot of different tasks there: looking for the necessary cables, people, setting up printers, fixing computers, bringing something in or out. We handled everything from cameras to rubber slippers.

We actively used our connections in the public sphere, found equipment and various necessary things from companies. During that time, we collected about a million dollars worth of equipment for the army.

There were military people in my family, and I have a military specialty. I was commissioned by the Military Qualification Commission and did not take part in combat operations before the start of the full-scale war, but I cannot call myself a civilian at all I understood how the army works – my «nannies» instead of grandmothers were military men.

From the very beginning, it was very problematic to process chaotic data

Every day the military came to the headquarters with different requests, but most of them had a common problem: the issue of processing chaotic data and transforming it into specific information.

This is a global problem, not just a Ukrainian one. For example, the story of the September 11, 2001 tragedy. The commission that analyzed the terrorist attack concluded that the intelligence community had pieces of data that would have allowed it to create a general picture and prevent this event. But they were in different places, in different offices, in different folders. And the intelligence services were unable to see the big picture in time to prevent the tragedy.

We had an idea to create a system that would process several streams of chaotic data, transform them into structured information and transfer them to military systems, for example, «Nettle», «Armor».

My command at the time agreed to try it. So I called my friend Oleksiy Teplukhin, who is one of the developers of «Armor», a very cool specialist. It was in early March 2022, and on March 5, we started working.

Oleksiy Teplukhin and Dmytro Shamrai at the National Defense Hackathon

It all started with a Telegram bot

Oleksiy brought his colleagues to join us It all started with a Telegram bot where civilians could send information. But we immediately faced a lot of questions: the information had to be systematized, checked, mapped, and passed on.

The issue of reliability was particularly acute. First, people themselves often confuse things and pass on incorrect information. Secondly, the enemy is constantly trying to disinform. In the first months of the invasion, for example, they often tried to direct our military to civilian targets in various ways.

To verify the reports, we need to collect other data about the place: what happened there yesterday, what happened a week ago? Were Russians there before? How many, when? Does the information from other sources corroborate? And so we gradually added new sources of information, the bot has grown into a whole system.

Dmytro Shamrai

The entire project team became members of the DFTG, initially there were five of us. We lived in the Kyiv City State Administration, which was also the headquarters. We slept on the floor, ate on the ground floor in the canteen, and showered there. We had two rooms, like offices.

At that time, it was loud in the capital, there were constant explosions, shots, and anxiety. All this interfered with work, and the guys often turned on gangsta rap.

When it came to the question of how we would name the system, the guys were like this: «O, Griselda, we are gansta».

When you have collected as much information as possible, you need to make the right decision

The scheme goes something like this: the system receives information that there is an occupant barracks in the village of Ivankove in the cultural center, and there is a lot of their equipment around. Then we collect more data: is this information confirmed by other sources? Does it match? Have Russians been there before? What is there around?” Information is collected from various types of data sources: satellite data, citizen data from the scene, open sources, aerial reconnaissance, hacked enemy databases, and electronic intelligence This data is partially processed using artificial intelligence.

Then we have to make a decision: do we have enough confidence that this information is true? This decision is ultimately made by the operator, our employee. Two such people work on a shift at the same time. This process often takes a few seconds. Moreover, a new operator can be trained in a few hours.

It is very important to make decisions with the right information. We saw this when the Russians decided to invade based on data that was misinterpreted. Everyone saw where this led.

No one had done this before, because there was no such big war

There are various analytical systems on the foreign market, but there is no such thing as Griselda The fact is that all armies always prepare for the last warbecause they study the experience of the past. It is impossible to predict how technologies will develop and how they will be combined. Remember that a year ago, everyone was quite skeptical about FPV drones, and now they are a powerful weapon.

We talked to NATO officers who work with information. Their systems have to answer questions, for example, with whom it is possible to interact in a particular village, who supports the rebels and who supports the occupiers, how influence is distributed among clans, what social ties are built between actors, who is the main actor in this territory, and so on. Or in the case of Afghanistan, to give some kind of economic picture In general, such systems structure information according to the PMESII ASCOPE matrix.

In conflicts, there was no such intensity and use of high technology on both sides. А We have a huge number of civilian technologies involved in the war – Telegram, Signal. And there are many civilians involved in the fight. No one could have predicted this. All of this data has to be combined into one overall picture to make decisions.

How the military uses Griselda

We mainly work with the J-2, J-5 and J-9 sectors. These are operations planning, intelligence, and civil-military cooperation management Our task is to simply collect, analyze and present the information to the military in a convenient form. We can adapt our «report» to most military systems, mostly Delta.

The information we collect can be different. It’s not just targets on a map: it can be a general description of the area. There are units that move frequently and need a convenient collection of data that will allow them to avoid wasting time familiarizing themselves with the area of responsibility.

It is important that we have no influence on the decision-making process of the military and do not know what has been decided. Our task is to develop and provide them with a convenient tool.

We currently have both civilians and military in our team. We understand that we have access to a huge asset of information, so we do not make any decisions and do not have access to these decisions. This is the task of the military, who have received information from us.

How the system helped save people after the Kakhovka HPP explosion

In June 2023, the Russians blew up the Kakhovka hydroelectric power station. Almost immediately, we were approached by local volunteers with the problem of processing the chaotic flow of rescue messages.

If someone is stuck on the roof of a flooded house, they will make all their friends write anywhere, call the State Emergency Service, ambulance, National Guard, police, write to all possible chats. And it becomes impossible to understand the real situation, to develop the logistics of rescuing citizens.

There may be 50 messages about the rescue of one person, and the person is gone – he or she has left. We were talking about duplicates, misinformation and prioritization, turning chaotic data into structured information for decision-making. These are exactly the same problems we have already solved for the military.

I hope that Griselda will become a business

The project is funded by donations and grants. We have used our experience in public activities to raise funds. Most of the project’s employees, and there are now several dozen of them, are volunteers. These are developers, operators, and people involved in attracting grants.

Griselda at the National defense hackaton

Currently, Griselda is a system that integrates other data collection systems, helps to use this information for decision-making, and creates a «bus» data to combine different military developments.

One of our main dreams is that one day we will grow into a common information space in which all Ukrainian defense entities will work.

Over time, I hope that we will become a fully-fledged socially responsible business. Our customers are already government agencies and we plan to expand this cooperation further. Thus, we will be able to strengthen government institutions with our products. I am sure that the future belongs to Ukrainian developments. After all is a concentrated combat experience that meets the real challenges of modern conflicts.

Author: MC.today journalist Natalia Solovyova.

Disqus Comments Loading...