Bob Pierce, PhD, is the co-founder and chief scientific officer of DiscipNext. His work brings advanced mathematical analysis into entirely new markets and industries, improving the way companies participate in strategic decision-making. Prior to decision-making, Bob was the chief scientist at SignalDemand, where he directed the science of the manufacturers behind its solutions. Bob has held senior R&D positions at Khimetrics (now SAP) and ConceptLabs, as well as academic positions at the National Academy of Sciences, Penn State University and UC Berkeley. His work covers a range of industries including commodities and manufacturing, and he contributes to econometrics, oceanography, mathematics, and nonlinear dynamics. He owns many patents and is the author of several peer-reviewed papers. Bob received his PhD in Theoretical Ph.D. from the University of California, Berkeley.
DecisionNext is a data analytics and forecasting company founded in 2015, specializing in AI-driven price and supply forecasting. The company was created to address the limitations of traditional “black box” prediction models that often lack transparency and actionable insights. By integrating AI and machine learning, DecisionNext provides businesses with visibility into the factors that influence their predictions, helping them make informed decisions based on market and business risks. Their platform is designed to improve the accuracy of the supply chain’s forecasts, enabling customers to go beyond intuitive-based decisions.
What was the initial idea or inspiration for building a decision? How did your background in theoretical physics and in various industries shape this vision?
My co-founder Mike Neal and I have gained a lot of experience in previous companies that provide optimization and forecasting solutions for retailers and commodity processors. Two main learnings obtained from this experience are:
- Users need to trust them to understand where the predictions and solutions come from; and
- It’s hard for users to separate what they think will happen, thus separating the possibility of it really happening.
These two concepts have deep origins in human cognition and the implications of how to create problem-solving software. As we all know, people’s minds are not good at calculating probability. As a physicist, I learned to create conceptual frameworks to interact with uncertainty and build distributed computing platforms to explore it. This is the technical foundation of our solution to help customers face uncertainty, which means they don’t know how the market will evolve, but still have to decide what to do now to maximize profits in the future.
How does your transition to the role of Chief Science Officer affect your daily priorities and long-term vision for decision-making?
The transition to CSO involves a renewed focus on how products should bring value to our customers. In the process, I have published some daily engineering responsibilities that are better handled by others. We always have a long list of features and ideas that can make the solution better, and this role gives me more time to research new and innovative approaches.
What are the unique challenges facing commodity markets that make them particularly suitable for adopting AI and machine learning solutions?
Modeled commodity markets present a fascinating combination of structure and random characteristics. The way to make it impossible to enter into a quantity of contracts with people for physical and paper transactions and the use of materials in production leads to a very rich and complex field. However, the development of mathematics is much better than the simple world of stocks. Artificial intelligence and machine learning help us work through this complexity by finding more effective modeling methods and helping our users navigate complex decisions.
How does decision-making Next balance the use of machine learning models with human expertise that is crucial to commodity decision-making?
Machine learning as a field is improving, but is still struggling with context and causality. Our experience is that human expertise and supervision are critical to the generation of powerful, minimalist models in some aspects of modeling. Our customers usually focus on the market through the perspective of supply and demand fundamentals. If the model does not reflect this belief (a model that is usually not supervised), then our customers generally do not develop trust. It is crucial that users do not integrate distrustful models into their daily decision-making process. Therefore, even a counterintuitively apparently accurate machine learning model will become a possibility on the shelf.
Client expertise is also crucial because the observed data is never completed, so the model represents a guide and should not be mistaken for reality. Users immersed in the market have important knowledge and insights that cannot be input as models. DecisionNext AI allows users to enhance model input and create market scenarios. This will enhance flexibility in predictions and decision recommendations, as well as enhance user confidence and interaction with the system.
Do you think there are specific breakthroughs in AI or data science that will revolutionize commodity predictions in the coming years and how do you position your own decisions for these changes nextext?
The emergence of functional LLM is a breakthrough that will take a long time to fully penetrate into the structure of business decisions. The pace of improvement in the model itself is still breathtaking and difficult to keep up. However, I think we are just starting to understand the best way to integrate AI into business processes. Most of the problems we encounter can be formed as optimization problems with complex constraints. Limitations in business processes are usually undocumented and not strictly enforced. I think this field is a huge opportunity for AI to both discover implicit limitations in historical data and to build and solve appropriate context optimization problems.
DecisionNext is a trusted platform that solves these problems and easily access critical information and predictions. DecisionNext is developing LLM-based agents to make the system easier to use and perform complex tasks in the system in the direction of users. This will enable us to scale and add value in more business processes and industries.
Your work covers the diverse areas of oceanography, econometrics, and nonlinear dynamics. How can these interdisciplinary insights help solve the problem of predicting commodities?
My diverse background informs my work in three ways. First, my breadth of work forbids me from diving into a specific field of mathematics. Instead, I have come into contact with many different disciplines that can be used to draw on all disciplines. Second, in all the work I do, high-performance distributed computing has always been a through line. Now, many of the techniques I use to piece together temporary computing clusters are used in mainstream frameworks, so I feel familiar even with the rapid pace of innovation. Finally, solving all these different problems will inspire philosophical curiosity. As a graduate student, I never intended to work in economics, but I am here. I don’t know what I’ll do in 5 years, but I know I’ll find it interesting.
DecisionNext emphasizes standing out from the predicted “black box” model. Why is this transparency so critical, and how do you think it affects user trust and adoption?
A prototype commodity trader (in or closing an exchange) is a skill that learns its industry basics in production but has betting skills in turbulent markets. If they don’t have real-world experience in supplying their business, they won’t win the trust of executives and won’t be promoted to traders. If they don’t have some affinity for gambling, they emphasize too much in executing transactions. Unlike the wise men on Wall Street, commodity traders usually have no formal probability and statistical background. To gain trust, we must propose an intuitive, fast, and touch on its cognitive biases, that supply and demand are the main drivers of large-scale market movements. So we took the “white box” approach where everything is transparent. Often, there is a “dating” phase where they are deep under the hood and we guide them through systematic reasoning. Once trust is established, the user does not take the time to go deeper, but will return regularly to ask for important or surprising predictions.
How does DecisionNext’s risk-aware prediction approach help companies not only respond to market conditions, but also actively shape their strategies?
Commodity trading is not limited to communication. Most companies only get future opportunities to hedge their risks. Processors may buy goods on the market as raw materials (perhaps cattle), but their output is also a variable commodity (beef), usually associated with the price of input. Given that expensive facilities must approach structural limitations of capacity, processors are forced to develop a strategic plan that can look to the future. That is, they cannot operate safely in the spot market completely, they must charge forward the contraction of their purchase materials and sales output. DecisionNext allows processors to predict the entire ecosystem of supply, demand, and price variables, and then simulate how business decisions are affected by market results. Paper trading may be part of the strategy, but most importantly, understanding materials and sales commitments and handling decisions to ensure capacity utilization. DecisionNext is tailor-made for this.
As someone with a deep scientific background, what makes you most excited about the intersection of science and artificial intelligence?
Behavioral economics has changed our understanding of how cognition affects business decisions. AI is changing how we use software tools to support human cognition and make better decisions. The efficiency gains that will be achieved by enabling automation by enabling AI have been widely discussed and will be economically important. Commodity companies operate with razor profit margins and high labor costs, so they may benefit greatly from automation. Beyond that, I believe most business decisions are made through intuition and rules of thumb, which is hidden inefficiency. Decisions are often based on limited and opaque information as well as simple spreadsheet tools. The most exciting result for me is for platforms like DecisionNext to help transform business processes using AI and simulations to standardize context and risk-aware decisions based on transparent data and open reasoning.
Thanks for your excellent interview, and readers who hope to learn more should visit DeciestNext.
