Content Authenticity Statement
100% of this article was generated by me, the human. I am the author of MaxBill Knowledge Hub for energy and utility decision-makers, LinkedIn Newsletters “Power Up with MaxBill”, and a keynote webinar speaker on E&U trends.
Artificial intelligence in energy and utilities looks risky because it’s difficult to anticipate ROI. It’s fair, as one expert at Digital Think Tanks of UK Utility Week once stated that ‘an overly enthusiastic or too liberal approach to exploring it could “bankrupt utilities”.
However, successful business cases of implementing AI prove this – AI is ‘the now’ for forward-thinking utilities in many ways, especially when it comes to productivity and efficiency.
Right before launching the MaxBill AI-billing and Product Catalog, we had a prospect (details are under NDA), a renewable energy supplier, that could not get the investor’s interest because of the lack of transparent, auditable financial operations.
So, they were the first to be reached out to give it a try with our next-gen solution – AI billing, supporting compliance, financial audits, and stakeholder reporting. As a result, they got their ops improved (we won’t dive into this part of case study) and the tech capacity to ensure a complete audit trail and billing transparency – automatically.
So, in this piece, we’ll talk about AI in energy and utilities but through a practical lens. This “lens” is the combination of
- MaxBill teams’ expertise in the field of energy and utility
- knowledge from consistent industry networking events
- collabs with partners that leverage AI to deliver extra value to the clients
- opinions of MaxBill R&D visionaries on what’s next for utilities.
Without further ado, let’s hit the road.
Artificial intelligence in energy and utilities: Artificial Intelligence Act
AI solutions in utilities are closely linked to data. In many use cases, to the customer one. So, MEPs and the Council have agreed on a legislative proposal to ensure that AI in Europe is safe, upholds fundamental rights and democracy, and enables businesses to thrive and grow. This is how the Artificial Intelligence Act was released.
The non-compliance may vary here: from 7.5 million euros, or 1.5% of turnover to 35 million euros or 7% of worldwide revenue. It all depends on the nature of the violation and the company’s size.
Now, what utilities cannot do according to the Artificial Intelligence Act?
- Categorise individuals based on sensitive characteristics like political, religious, or philosophical beliefs, sexual orientation, and race,
- Scrape untargetedly facial images from the internet or CCTV footage to create facial recognition databases,
- Use AI for emotion recognition in workplaces and educational institutions
- Assign scores to individuals based on their social behaviour or personal characteristics,
- Manipulate human behavior in a way that circumvents free will,
- Exploit the vulnerabilities of individuals based on age, disability, and social or economic situation.
In this ‘carrot-and-stick’ approach, where’s the carrot? Utilities are encouraged to:
- Join regulatory sandboxes that represent controlled environments where businesses, especially SMEs, can test and develop innovative AI solutions. They allow for experimentation and innovation without the immediate pressures of full regulatory compliance.
- Participate in real-world testing environments, facilitated by national authorities, that enable SMEs to train and refine their AI systems in practical settings. They say that this hands-on experience helps businesses optimize their AI applications and bring well-tested products to market.
- Collaborate with other businesses, research institutions, and regulatory bodies. This leads to shared knowledge, resources, and advancements in AI technology, benefiting the entire industry (like we do with Tillix, for example).
Additionally, the Act aims to prevent undue pressure from industry giants by creating a level playing field. This includes measures to ensure that SMEs have access to the necessary resources and support to develop their AI solutions independently.
Now, let’s get down to real-life business cases of AI implementation in energy and utilities.
Real-life business cases of implementing AI in energy and utility
The veracity of the output of AI is growing over time and will continue to do so. It’s complicated to categorise the application of AI in E&U because it’s literally everywhere.
Here we’ll outline the following area:
- Customer service and engagement
- Billing, tariffs, and financial optimisation
- Energy demand, load balancing and grid optimisation
- Infrastructure and asset management
- Decarbonization and energy transition.
Customer service and engagement
AI-powered solutions for customer service are literally a major productivity booster for customer-facing teams. The following are strong examples of this:
AI-powered customer service automation
AI-powered chatbots address high-bill queries, reducing the need for human intervention. It analyses historical usage and billing information to provide real-time responses to inquiries.
AI call summarization and live transcripts
AI-based summarization tools allow agents to transcribe customer service calls, generate conversation summaries, and draft automated email responses. This way, they improve response times and operational efficiency.
Automating inbound email handling
AI allows to automate inbound email handling for technical inquiries. Utilizing AI-driven assistants in core applications like Outlook and Teams adds to CSRs’ productivity.
Interpreting intent to help overloaded call centers
Nowadays, Large Language Models interpret intent from inbound customer inquiries via chatbots and emails, which streamlines triage for agents.
Whatever your utility company’s profile, these models can be trained with other company-specific datasets, such as customer attrition, debt tendencies, and solutions to common exceptions to reduce costs.
Automating feedback analysis
A transmission operator (TO) uses AI to process public feedback on projects. AI analyzes stakeholder responses. As a result, savings are over 80% of time, cutting down analysis from days and weeks to minutes and hours.
Identifying vulnerable customers
Utilities are exploring AI-driven predictive maintenance to pinpoint individuals who may struggle with digital tools and are at risk of digital exclusion. This way, they ensure these customers receive the necessary support.
Billing, Tariffs, and Financial Optimisation
Analyzing revenue-related data
AI-driven systems analyse historical billing data, compare tariffs, and identify discrepancies in invoices for businesses with large-scale energy consumption. AI also forecasts market-driven tariff fluctuations to optimize costs.
Addressing customer attrition and debt-risk contracts
The forecast models tackle customer attrition and the potential occurrence of debt-risk contracts. With the ‘what-if simulation”, the models generate suggestions about this or that offering for this or that particular client to make them stay and keep using the service.
Energy demand, load balancing and grid optimisation
Energy demand forecasting and load balancing
Machine learning models process historical energy consumption data, weather patterns, and regulatory influences to predict future energy demands. AI enables better load balancing, prevents grid overloading, and improves demand-side response strategies. This leads to reduced energy waste and improved energy efficiency.
HVAC and building energy optimization
AI enhances HVAC efficiency by analyzing consumption trends, weather data, and occupancy patterns to adjust heating, ventilation, and cooling settings dynamically. AI ensures that energy is used only when necessary. It automates system adjustments and prevents excessive heating or cooling, which results in substantial energy savings.
Strategic data utilisation
Sorting all the data could take decades. That’s why strategic selection of high-value datasets powered by AI is absolutely a way out. Now, this approach is used within transmission operators. AI identifies the most relevant datasets for specific applications. This allows for the avoidance of unnecessary costs associated with indiscriminate data storage.
Infrastructure and asset management
Identifying high-risk sewage pumping stations
It analyses data on pump behaviour, power consumption, wet well levels, and rain impacts. Field trials revealed that proactive inspections based on this model prevent about one pollution event per month.
Predicting the root causes of failures on the water network
ML models predict water demand by analyzing sensor data from reservoirs, treatment works, and water supply networks combined with hyper-precise weather data. The model, which refreshes daily and visualises predictions on a dashboard, achieves 98% accuracy. This aids operations teams in efficiently managing storage and preparing for demand peaks.
AI for project management
A transmission operator uses AI to predict project outcomes. AI compares historical data on past projects with current schedules to assess deliverability and identify risks. The tool highlights alternative critical paths and predicts delays. Thus, new project managers plan more effectively.
Decarbonization and energy transition
Introducing flexibility services
AI streamlines the planning of overhead lines and simplifies the installation of heat pumps by automating assessments and calculating savings from retrofitting. This underscores its immense value in driving decarbonisation efforts.
AI in revenue cycle management: MaxBill AI billing
MaxBill has released the AI billing and product catalog to turn complex pricing models into product catalogs and create portfolios – literally in minutes. The same goes for charging – in a matter of clicks. What earlier used to take weeks, now is supposed to take minutes.
The good news is that there’s no need for the implementation project, as it is available here and now.
So, what do AI algorithms do?
- Interpreting contracts or plain text to automatically generate billing structures, which reduces setup time by up to 90%.
- Supporting rapid deployment of new features and services, which facilitates expansion into new markets with minimal setup.
- Adapting to evolving rules, pricing, and billing cycles without manual updates, which ensures seamless compliance and responsiveness.
- Automation leads to approximately 30% savings in ops by minimizing manual intervention.
- Providing insights into optimal offers, which potentially results in a 10–15% increase in revenue through better-targeted services.
- Supporting multi-tenancy, multiple languages, and currencies, with real-time process optimization through large language model agents.
In the context of the whole AI revenue cycle management, the brightest sides of AI billing are automation and transparency. The latter ensures visibility and clarity of revenue sharing and financial settlements, which ensures revenue leakage prevention and healthy case flow, in the first place.
How E&U CIOs should start implementing AI in energy and utility
Let us define the top 7 universal recommendations that utility CIOs will definitely find helpful:
- Focus on AI projects with clear business value and preferably start with low-risk, high-impact use cases. These might be forecasting, document triage, or admin automation.
- Large proof-of-concept efforts are gone. It’s risky and expensive. Therefore, use minimum viable products (MVPs). Plus, it’s more optimal to pilot in one area, then expand based on results.
- Data often flows unprepared for further use. Establish data governance and eliminate internal silos and then monitor its veracity, reliability, and relevance.
- Before implementing AI, redesign processes and workflows. Teams can start with operational areas (like asset performance or customer experience) where automation and prediction offer immediate efficiency gains. It’s important to notice that force majeure situations occur and algorithms might not be trained for certain ‘fires’ in the company. So, human supervision is still necessary.
- Building internal skills should be one of the company’s priorities. External vendors will definitely help out, but it’s better to reduce such reliance over time by building internal data science and AI teams.
- Have a clear strategy and governance framework around AI, aligned with business goals and appetite. The AI implementation roadmap has to include AI ethics, legal, and cybersecurity guardrails from the start.
- Teams of any level should recognise: AI is “creeping” into every tool and doing nothing is riskier. Competitors are already leveraging AI to deliver greater service.
In the meantime, the AI application is not a one-time transformation. So, utilities should set expectations realistically and iterate AI solutions, keeping in mind a long game.
