How Artificial Intelligence Ensures Safety in Oil and Gas Operations: An Interview with Andrey Bolshakov from NVI Solutions

Andrey Bolshakov, co-founder of NVI Solutions, discusses how his company is transforming safety practices in the oil and gas industry through AI-powered digital supervision systems. He also shares his insights into the differences in AI adoption across Asia and Europe.

Andrey, to start, can you tell us about NVI Solutions and your role in the company?

At NVI Solutions, we provide AI-based digital supervision services. Our clients are major companies managing critical infrastructure, such as oil refineries and gas production facilities. These facilities house complex and costly assets, which are often serviced by various subcontractors. Unfortunately, these subcontractors don’t always adhere to operational regulations, sometimes bypassing technological protocols or mandatory procedures. This can result in equipment damage or even worker fatalities. Our solutions are designed to prevent such outcomes.

NVI Solutions assists our clients in enforcing regulatory compliance. We digitize technical guidelines and create operational scenarios for our software. We then install systems comprising smart cameras and various alert mechanisms at the facilities. These systems integrate with the facility’s infrastructure to receive telemetry—an independent stream of technological data that helps identify ongoing operations.

Using the telemetry and predefined scenarios, our software evaluates factors such as the number of personnel, their attire, actions, and the condition of equipment. If there is a critical deviation from the regulations, our system signals that the operation is proceeding incorrectly. For instance, if a contractor promises to send 100 workers but only 30 arrive, or if smoke appears on-site due to worker negligence, our system detects and flags these issues in real time.

The alert system operates on three levels:

1. On-site alerts: Workers are notified via loudspeakers in their native language.
   
2. Supervisor notifications: Supervisors and management receive immediate alerts about the violation.
   
3. Escalation: Further actions are taken, depending on the severity of the situation.
   

This multi-tiered signaling system ensures quick, effective responses to potential hazards or operational lapses, improving safety and regulatory compliance across complex industrial sites.

I’ve always been passionate about creating systems that help people. Our core mission—saving workers’ lives—is deeply personal and motivating for me. About 4–5 years ago, I met my future business partner, Philipp, who is now the CEO of NVI Solutions. Philipp had extensive experience in sales within the oil and gas industry and understood which investments were worth pursuing and the key challenges in this sector. Combining my expertise in technology and product management with his knowledge of sales and economics, we founded NVI Solutions.

At NVI, I’ve been responsible for product strategy from the beginning—defining the product’s features, ensuring it supports the right technologies, fits within budget constraints, and aligns with long-term economic goals. In addition to strategy, I oversee the technological aspects of the product, including improving monitoring systems, adding supplementary equipment, and integrating new technologies. Philipp handles sales and client relations. 

These tasks are particularly challenging because we operate in high-tech environments where thousands of operations occur, and even minor errors can be extremely costly. Our software and equipment must work seamlessly with the AI to ensure reliable performance.

You participated in ADIPEC 2024 earlier this November. What key ideas and technologies did your company present at the exhibition?

ADIPEC is one of the largest global exhibitions focused on oil and gas technologies. While I’ve attended it several times as a visitor, this year marked our first time presenting our solutions. Previously, we participated in international IT exhibitions like GITEX and InterSEC. However, ADIPEC directly aligns with the industry we operate in, so we were excited to showcase our product there.

The feedback we received was overwhelmingly positive. Many companies are increasingly prioritizing the safety of their employees and technological processes. A growing number are either developing or seeking to acquire systems like ours, recognizing the vast potential of automation. On-site supervisors often cannot monitor all operations effectively—when they visit, workers might comply with regulations and wear uniforms, but once the supervisors leave, the situation becomes uncertain. AI, on the other hand, can continuously analyze operations 24/7 and flag any violations.

We noticed significant interest in our product, especially from some of the world’s largest corporations, such as ADNOC, the leading oil producer in the UAE, and Saudi Aramco, the national oil company of Saudi Arabia. We had in-depth discussions with these and other key players, outlining potential next steps for collaboration. With some, we even signed memorandums of understanding to formalize future partnerships.

Moreover, we’ve already initiated pilot projects at several Saudi Arabian oil refineries and in the ports of UAE. In UAE, we’re also exploring opportunities to implement our solutions in the security services sector.

What sets us apart is our willingness to take on challenging environments. Unlike many startups that operate comfortably from business center offices, offering their products remotely, we’re ready to go the extra mile—whether to the Arctic, scorching deserts, or other harsh conditions—to implement and test our systems on-site. This adaptability has become a defining strength of our company.

Let’s talk more about your solutions for the oil and gas industry. How do you assist businesses in this sector?

Even in industries like oil and gas, violations are common. For instance, workers sometimes skip safety protocols to complete tasks more quickly. Contractors might send unqualified personnel and stretch out work to earn more money for the extra time spent. Without proper oversight, employees often fail to follow regulations, as this doesn’t seem to result in immediate consequences. However, over time, this leads to incidents and financial losses.

Our system helps minimize these issues. At the same time, it enhances work culture and efficiency because, unsurprisingly, when cameras are present on-site, employees are far less likely to break the rules. Our solution also strengthens subcontractor oversight, ensuring they don’t try to cut corners and that they send the correct number of people as outlined in the contract.

What auxiliary innovations in video data analysis are driving value for your clients in the banking, agricultural, and logistics sectors?

One promising area is Optical Character Recognition (OCR) technology, which is designed to extract data from standardized documents. These might include construction documentation, work orders, and other types of paperwork that are often only available in physical form, and sometimes handwritten. We can digitize and recognize these documents, transforming them into multimedia documents that our AI systems can use for decision-making.

In banking and logistics, a significant amount of paperwork still exists in physical form. Our tools can quickly recognize these documents while simultaneously verifying their authenticity. This drastically reduces operational costs, such as those associated with managing banking credit workflows. AI also speeds up operations: instead of waiting for an operator to enter information from an invoice into the system manually, documents can be quickly scanned and sent directly to the client.

We also see great potential in the agricultural sector. Despite the high volume of processes and human labor involved, many tasks are still not digitized. For example, in agriculture, having an ERP system implemented is considered a major achievement. In regions with less developed agricultural processes, average crop yields are often half of what they are in Europe, and I believe that enhanced oversight could improve results by 30-50%.

How do you see the future of video analytics integration in other business segments? Are there industries you’re planning to enter?

My personal immediate focus is the agricultural sector. I see many similarities with the industries I’ve worked in before, but the potential for digitalization in agriculture remains largely untapped. I would leverage the expertise from both of my companies — NVI Solutions and Megawatt — in this area. For example, we could use mobile cameras mounted on unmanned vehicles to provide surveillance across large areas.

Although the agricultural market doesn’t have as much concentrated capital as the oil and gas industry, it has a vast number of players, which creates great opportunities for growth. Additionally, I’m extremely interested in working with large-scale business models that span thousands of hectares. For me, this sounds like a real challenge.

Tell us about your R&D center and its key developments.

Our primary devices are smart cameras. The video-processing system is installed directly on the site. This is quite unique because most companies upload their data to the cloud, but our sites are often located in remote areas where there’s no internet access. Moreover, we work with critical infrastructure, and clients prefer to keep their video streams within secure, local systems rather than sending them to the cloud. Therefore, our R&D center is focused on efficient real-time video stream processing. This needs to be done quickly and within the most secure environment possible.

Additionally, we have a very robust infrastructure for generating synthetic datasets. When developing AI systems, we often face the challenge of finding enough data to train the models. Every incident is unique, and it’s impossible to collect thousands of necessary examples. That’s why we’ve developed an engine that allows us to simulate objects with photorealistic accuracy and “create” potential incidents with them. Essentially, it’s a simulation similar to a high-quality video game, which shows what happens to an object and its people before an incident occurs. Our developments in this area are unique.

What trends do you see emerging in video analytics in the coming years?

I see several powerful trends emerging. The first is analytics directly within the cameras, also known as edge computing. Developers are realizing that there’s no need to send data to a data center; operations can be performed on-site instead.

The second trend is more complex detections that intersect with systems like ChatGPT. This involves cameras or equipment setups attempting to “understand” and “explain” what’s happening at a site, without relying on standard scenarios.

There’s also a big push for systems to become more comprehensive. More and more sensors are being integrated, and this data is fed into a Data Lake, creating vast amounts of data that can be used to predict numerous incidents.

Another key development is the creation of so-called digital twins for businesses. Right now, there are solutions that some call digital twins, but essentially, they are just mirrors of the enterprise—visualizations of operations. A true digital twin replicates all objects and personnel, displaying all production data, and allowing businesses to analyze what will happen to their operations over the next week, month, or year. This data can then inform strategic decisions through management systems, creating an ability to influence the physical world. Currently, there are no tools on the market capable of creating such comprehensive digital twins.

What is your personal take on the AI and digital twin developments?

I favor an evolutionary approach. I feel that in some areas of technology, we’ve advanced so rapidly that we’re struggling to fully grasp the world we now live in and what lies ahead. I believe we need to understand the scale of change a technology will bring and, based on that, introduce it while carefully managing the associated risks.

That said, there are different approaches to this issue. I encountered this firsthand about five years ago when I was working simultaneously with China and Western Europe. Clients in China were ready to implement our products right away, whereas in Europe, I had to prove that such systems would benefit individuals without intruding too much into their personal space. This reflects fundamentally different approaches to designing business value systems.

This contrast leads to an interesting story. At the same time, Europe has implemented many regulations around artificial intelligence, Asian countries, especially China, have not. As a result, Chinese companies have made far more progress, while Europe has fallen technologically behind by decades. Now, Europeans face a choice: either they can isolate themselves within their own ecosystem or allow Chinese businesses to enter their markets, bringing different values and a culturally distinct economic framework.