(Translated with ChatGPT without further editing)
Today, we welcome Innokentij Bogatykh to our Hot Seat as a representative of the three-person founding team of ModelForge. The Augsburg-based start-up is revolutionizing AI-driven process optimization for the chemical industry. We talk with the team about their mission, the challenges facing the industry, and how their technology addresses them.
Don’t miss ModelForge’s upcoming Spotlight webinar! 🗓️03.02 ⌚ 11:00, 📍online! Register now:
Hello to the ModelForge team. Could you give us an overview of ModelForge and its mission?
Operating chemical plants in an economical and stable manner is challenging, as decisions often have to be made under time pressure and with incomplete information in complex, dynamic systems. ModelForge builds on years of research in process optimization and uses AI to reduce cognitive load for plant personnel and operate facilities more efficiently. To achieve this, we have developed a platform that enables the rapid and tailored creation and integration of AI-based assistance systems for specific plants and concrete problems.
What specific challenges in the chemical industry does your software solution address?
We address the reality of plant operations: processes are constantly changing. Classical predictive models used for optimization quickly lose performance under these conditions, and systems degrade over time. As a result, model-based solutions often become expensive, maintenance-intensive one-off projects with high effort. Our solution is adaptive process models that automatically adjust to changes and are easy to create. This keeps system performance stable while significantly lowering the barrier to entry.
How does the AI-based process optimization in your software work, and what advantages does it offer compared to conventional methods?
AI-based process optimization uses artificial intelligence to predict how a process will evolve. This allows both undesired and desired operating states to be identified early, and the system recommends interventions that move operations closer to the economic optimum. Humans always remain in control. Compared to classical methods, the solution can be quickly adapted to a wide range of chemical processes and plants and requires no ongoing maintenance. In addition, it is flexible enough to be transferred to similar plants without noticeable loss of performance. This transforms software-supported process optimization from an expensive one-off project into a scalable fleet-wide solution.
Can you give examples of successful applications of your tools in industry?
We achieve the best results in batch processes. Here, initial conditions vary significantly from batch to batch, and decisions often have to be made under time pressure. These processes include batch reactions and batch distillations. The optimal operating strategy cannot simply be derived from fixed recipes. The AI predicts the batch progression at an early stage, makes deviations and critical states visible in time, and provides concrete recommendations for operation. As a result, batches run faster and more stably, and operators are relieved.
Dr. Innokentij Bogatykh on himself and ModelForge:
During my doctoral research, I observed that strong research approaches often fail to reach plant operations because integration and reliable operation are too complex and resource-intensive. These topics are rarely addressed systematically—not because they are unimportant, but because they offer little potential for academic publication. Whether software crashes less frequently, integrates more cleanly, or becomes stably productive within a short time is difficult to capitalize on scientifically. In practice, however, it is precisely these factors that often make the decisive difference.
That is why my co-founders and I founded ModelForge: to make state-of-the-art AI methods available in a way that is robust in operation, quick to integrate, and delivers clear economic value.
