How to Gain a Competitive Edge with Generative AI in Energy and Commodities

Organizations can expect an immediate material impact on business outcomes by deploying generative AI through the augmentation of corporate tasks, improvement in risk management and optimization of core strategic capabilities.
Energy and commodities companies can expect to realize quick wins by deploying generative AI to accelerate routine tasks and codify institutional knowledge.
To effectively manage change, early adopters should focus on appropriate governance and guardrails by quickly defining proof of concepts and ensuring data readiness and pipelines.

The generative AI revolution is here.

Energy and commodities sectors are on the precipice of a major transformation in ways of working and delivering value to customers. By using generative AI in tandem with existing tools, organizations can gain a competitive advantage. To do this, they should address several key questions:

• What does generative AI do?
• How can energy and commodities organizations leverage generative AI to move faster, save money and better serve customers?
• What are the implications for the energy and commodities workforce?
• What potential risks, liabilities and regulations should organizations consider?
• What can organizations do in the short, medium and long term?

1. So, what can generative AI actually do?

The adoption of AI has dramatically accelerated in recent years and will only continue to increase as consumer-friendly generative AI tools reduce the barriers to AI adoption. AI has historically been implemented for most manufacturing operations as a means to reduce operational costs, improve reliability (and thereby reduce capital expenditure) and increase yield. Consider several applications already being deployed across energy and commodities sectors:

• Oil & Gas 4.0: Applications for AI in the energy sector were first deployed in the 1970s, and these include upstream natural gas fracking, midstream pipeline monitoring and downstream autonomous oil refining (e.g., Fluidized Catalytic Cracking)
• Smart metering and detection: Utilities enable smart metering and proactive anomaly detection within transmission and distribution infrastructure
• Agricultural monitoring: Farmers employ real-time monitoring to surveil their crop yields and livestock

Generative AI: making sense out of context

Specifically, generative AI can create contextualized content—ranging from code and long-form writing to images and videos—based on data and human-engineered inputs, simulating human language with reasonable accuracy. Furthermore, as a standalone tool, generative AI can significantly enhance workforce productivity by structuring disparate and siloed data and translating multilingual text. Considering its pace of development, potential generative AI tools will rapidly become more efficient and autonomous in managing workforce tasks. Moreover, generative AI’s ability to be interoperable with existing solutions—including machine learning (ML), cognitive automation, internet of things, robotic process automation, optical character recognition and natural language processing—can unlock differentiated operational value. Using natural, human-like language to interface and assist with other layers in the AI “tech stack,” generative AI can enable early and agile AI adopters to:

• Autonomously execute operational tasks—such as asset optimization and trading risk management—with provided logic and algorithm-based rules
• Lead self-supervised learning to predict and generate content
• Institutionalize organizational knowledge and continuously learn new information
• Generate creative ideas and solutions

2. How can energy and commodities organizations leverage generative AI to enhance existing tools and processes?

Energy and commodities participants can expect accelerated productivity through democratized access to complex digital tools and a rapid iteration cycle to ideate, design, code and communicate. The use of large language models can be a powerful differentiator when coupled with proprietary information and connected to existing technology ecosystems. While most organizations employ data lakes, predictive analytics and machine learning, generative AI adds a powerful interactive layer on the top of this AI tech stack that unlocks value for operations, asset optimization and risk management across energy and commodities sectors. Generative AI paired with existing tools has the potential to unlock new value for energy and commodities organizations in six ways:

1. Augmented supply and trading for commodities traders
   Benefit: Helps commodities traders make more informed decisions about trading and hedging commodities.
   Use cases
   • Real-time market monitoring: Monitor commodity markets to identify critical supply and demand trends
   • Demand forecasting: Analyze historical commodity demand curves and external macro factors to generate demand forecasts
   • Price forecasting: Build predictive models to forecast commodities futures curves, informed by historical pricing trends, futures prices, volatility, macroeconomics, market sentiment and other external factors
   • Logistics and supply chain optimization: Analyze historical datasets on transportation routes, storage capacity and demand patterns to optimize commodity transportation and storage
   • Risk management: Generate synthetic scenarios to assess the market, credit and liquidity risk on a company's commodity portfolio and propose potential hedges that can be automatically executed. Develop risk management strategies through assessment of risk policy, identification of policy violations and design of potential hedges
   • Portfolio optimization: Define and identify the “efficient frontier,” i.e., the most profitable commodity combination to optimize commodity portfolios
2. Proactive and predictive maintenance for operators and field technicians in downstream refining and chemical processing units
   Benefit: Equips operators and technicians to devise optimized maintenance schedules and prevent system failures.
   Use cases
   • Reporting: Provide stakeholders with relevant asset condition data based on role and responsibilities. For example, maintenance teams receive key inputs on asset failures and potential corrective actions; sales and operations planning teams receive synthesized data and insights on potential supply impact and mitigation actions to meet customer demand
   • Semantic search and pattern detection: Identify patterns of asset faults and corrective actions by querying training manuals, historical maintenance notes and best practices from industry or to consolidate knowledge on asset performance and downtime
   • Tactical recommendations: Automated generation of remedial instructions and communication to field agents based on results of historical asset fault detection and maintenance
3. Refinery process optimization for operators, engineers, and sales and operations planning teams in downstream refining and high-volume chemical processing
   Benefit: Enables organizations to more efficiently manage and monitor supply chains and emissions.
   Use cases
   • Asset management and monitoring: Generate operation and compliance logs, optimize maintenance schedules and improve efficiency to reduce downtime and costs
   • Crude assay and feedstock selection: Autonomously adjust operating conditions, catalyst formulations and process configurations to maximize yield and quality and limit capital-intensive changes to infrastructure
   • Value chain optimization: Analyze inventory levels, shipping schedules and market demand to optimize system-level supply chains, simulate scenarios and identify cost-saving opportunities
   • Inventory management: Synthesize real-time inventory levels, market demand and production schedules to optimize the inventory levels and reduce inventory holding costs
   • Energy optimization: Drive cost savings and reduce greenhouse gas emissions with load balance of energy inputs and outputs
   • Environmental monitoring: Collate disparate data sets to improve real-time monitoring, quantify emissions, reduce overall environmental impact and generate compliance reports for regulatory agencies
4. Customer engagement for retail power and utilities companies
   Benefit: Empowers organizations to elevate customer engagement and gain deeper insights into customer behavior, expectations and needs.
   Use cases
   • Call center analytics: Extract and synthesize salient information and key themes from customer call logs and complaint forms, and generate customer communication through multiple forms, including chat, email, text, voice or video. Deploying generative AI can more effectively support issue triage and empower human agents to make faster decisions and improve customer response times
   • Pattern recognition: Synthesize, classify, organize and analyze customer conversations to identify emerging patterns and preemptively address issues of increasing importance. For example, identify increasing consumer complaints at specific times of day and better match load, infrastructure and maintenance response to meet power demand
   • Hyper-personalization: Design custom recommendations for customers related to power consumption strategies, adopting renewable technology behind the meter and promoting sustainable energy management
5. Monitoring livestock and crops for agricultural product manufacturers
   Benefit: Reveals real-time analytics and insights that empower farmers with knowledge.
   Use cases
   • Create training data: Create sample data sets about livestock and crop fields to test for abnormalities; pair with real-time environmental data (i.e., weather patterns, crop trade prices), simulate business outcomes and develop more accurate business cases to procure capital equipment
   • Contextualize large data sets: Synthesize and hypothesize trends—such as weather patterns, product pricing, crop and livestock health, and logistics availability—to quickly gain preliminary insights without depending on external industry “trusted advisors”
   • Design remedial treatment programs: Leverage databases and real-time data to personalize corrective programs and draft necessary regulatory documentations to remain in compliance with governing agencies. For example, once images of crops and livestock have been labeled, identified and analyzed, generative AI can design treatment regimens and recommend preferred vendors that can quickly supply the desired restorative product. Furthermore, generative AI can draft necessary regulatory documentations to remain in compliance with government agencies
   • Write code: Accelerate the writing of programming code that will integrate with existing AI- or ML-powered tools; e.g., write a program to adjust temperature for an indoor vertical farm
6. Strategies for industry regulatory agencies governing each sector
   Benefit: Ensures that agencies can efficiently perform and complete inspections.
   Use cases
   • Safety inspections: Accelerate pipeline safety inspection programs and audits to more efficiently use government funding
   • Alleviate ISO/RTO interconnection queues: Sort and expedite the highest impact projects through automated application processing, supply and load simulation and decision generation
   • Manage backlogs: Rapidly process farmer complaints, generate commodity reports and communicate small farm loan issuance decisions to maintain food safety, quality and supply

3. How will this impact the energy and commodities workforce?

Generative AI will transform the connected worker experience and thereby improve workforce efficiency, particularly for corporate functions such as sales, marketing and financial operations. By Publicis Sapient’s estimate, generative AI may improve the efficiency of tactical back-office activities and reduce approximately 10-30 percent of corporate costs through automation of tasks like data cleansing, data validation, research/planning and drafting. These core tasks can augment or displace certain corporate functional activities, significantly reducing the number of labor hours required. Displaced jobs may be offset by new technology-driven roles, or “AI Complements,” that focus on strategic value-creation activities such as designing innovative crop cultivation or troubleshooting refineries. The resulting productivity gains could increase GDP by seven percent over 10 years.

How will generative AI help mitigate energy and commodities workforce attrition?

More specifically, generative AI can help mitigate workforce attrition and the resulting brain drain within the industry by codifying and institutionalizing existing knowledge and organizational best practices. Over the next decade, energy and commodities sectors will face a significant portion of the workforce retiring and aging. For example, approximately 27 percent of oil and gas laborers are currently over the age of 55, and the age of agricultural laborers increased by eight percent since 2018. As energy and commodities sectors continue to manage this ongoing brain drain, generative AI can help upskill early-career professionals and therefore reduce the learning curve and accelerate an individual’s ability to generate value for the organization.

4. What potential risks, regulations and reactive measures might organizations face?

Generative AI has clear operational and strategic potential for energy and commodities organizations ready to make the digital leap. However, implementing effective governance measures around data use and managing outputs is critical to avoid unintentional consequences. Consumer-facing applications (e.g., ChatGPT) leverage data from the open web. As a result, organizations may be at risk of proprietary data leakage if users prompt with company-specific data. Furthermore, organizations should be aware of functional limitations when deploying generative AI tools:

• Large training data: Large baseline data sets are required to effectively replicate and produce language with reasonable accuracy
• Knowledge cutoff: Some tools have a knowledge cut-off date
• Low maturity: Currently limited to question-answer format
• Superficial search function: Requires industry-level and company-level knowledge to make it relevant
• Limited quantitative reasoning ability: Must be paired with a machine learning algorithm or another analytical approach to be effective
• Misinformation (or “hallucination”): May provide inaccurate or “fake” information in a confident manner
• Response ambiguity: Clear questions and prompts are required to provide precise answers and, as a result, makes it susceptible to data poisoning
• Ethical concerns: Risk of generating harmful, biased or unoriginal content
• Sensitivity: May not understand user connotation or emotions elicited by responses

Generative AI has also been a topic for regulatory bodies around the world. The European Union’s Artificial Intelligence Act, for example, seeks to regulate AI in order to mitigate risks. To responsibly adopt generative AI technologies, leaders must understand the risks, potential regulatory reactions and limitations inherent to generative AI.

5. What can organizations do to ignite a successful, well-defined and collaborative roadmap?

Once an organization has fully explored questions for generative AI, they can start building their roadmap for success by focusing on what it can do today, tomorrow and down the road:

Do now

• Create a shared knowledge base of the capabilities and limitations of generative AI tools and improve transparency within the workforce to demonstrate the potential for digital upskilling. It is critical that all stakeholders understand the boundaries of generative AI from other tools such as ML or predictive analytics
• Test generative AI within your existing capabilities to identify a long list of use cases specific to the organization. Brainstorm with operations and transformation teams on value-generating activities that can be accelerated, displaced or unlocked by integrating generative AI with existing autonomous tools
• Quantify the size of the value pool and determine the complexity of implementation to prioritize “quick wins” across the organization

Do soon

• Establish clear data governance, security, compliance and best practices around the enterprise use of generative AI in your organization through cross-functional and cross-business-unit collaboration. Set guardrails around the appropriate use of outputs created from generative AI and prevent data leakage
• Define proof of concepts, gather resourcing and technical requirements, and detail business impacts to enable organizational and operational change
• Design proof of concepts to demonstrate applicability and outcomes without the significant capital outlay of full-scale implementation

Plan for the future

• Set up your ML platform environment. Activities will also include configuring data preparation, training and production resources, infrastructure, workflows and tools. Institute feedback mechanism through continuous monitoring and operation of the ML/generative AI platform. If value proposition declines over time, identify requirements to retain, remodel or reuse to ensure continuous value creation.

Generative AI presents an opportunity for energy and commodities organizations to improve efficiency and unlock value by adopting new tools and optimizing the ones they currently use. By understanding generative AI’s power and potential today, organizations can start building a more profitable tomorrow.

As part of its digital transformation plan, Publicis Sapient helps organizations leverage the power of generative AI to gain a competitive advantage. Reach out today to learn how your organization can take part in the generative AI revolution.

Aash Jain

Gen AI Energy Vertical Lead, North America
Let's connect

Sachin Chanana

Senior Director, Product Management
Let's connect

Nitin Gulati

Director Technology
Let's connect

Tom Peffers

Director Business Development
Let's connect