ARTICLE: How to Identify High-Impact AI Use Cases for Your Business

In the first two parts of this series, we’ve explored the fundamentals of Artificial Intelligence (AI) and the practical steps you should take to be ready to implement it within your organisation.

Now, it’s time to consider some live examples to move from the building blocks to where the opportunity exists and is delivering measurable results.



Octopus Energy pioneered an innovative approach to customer service by leveraging the combined power of Custom Language Models (CLMs), Automatic Speech Recognition (ASR), and Sentiment Analysis. Here’s a breakdown of their impressive system:

Custom Language Model (CLM): They trained a CLM specifically for the energy industry and their company’s unique terminology. This ensures the AI model can accurately understand the nuances of energy-related conversations, significantly improving speech recognition accuracy.

Amazon Transcribe: They use Amazon Transcribe, a cloud-based ASR service, to capture and convert the audio from all their customer calls into text.

Kraken CRM System Integration: The call transcripts are then fed into their CRM system, Kraken, every 15 minutes. This provides rapid access to a complete customer history for their call agents.

The power of this system goes beyond simply providing transcripts. Here are some additional advantages:

AI-powered Summaries: The AI can automatically generate a summary of each call, allowing agents to quickly grasp the key points and customer concerns before interacting with them. This improves call efficiency and ensures agents are well-prepared to address customer needs.

Sentiment Analysis: The system can analyse the sentiment of the conversation, identifying frustrated or dissatisfied customers. This allows Supervisors to prioritise calls and ensure prompt intervention for those in need.  


Octopus Energy reported a 21% increase in transcription accuracy thanks to the custom-trained CLM. More importantly, they’ve observed a significant improvement in faster resolution times for customer enquiries.

Assessing Money Laundering Risk – HSBC

Partnering with Google Cloud, HSBC developed a Dynamic Risk Assessment (DRA) system to identify suspicious activity more effectively.

This system can analyse vast amounts of financial data, spotting hidden patterns that might escape the human eye. Here’s a glimpse into how it works:

Data Powerhouse: The AI is trained on a massive dataset of real-world financial information, including Know Your Customer (KYC) data and transaction patterns. This rich data pool allows the AI to identify anomalies and suspicious behaviours.

Staying Sharp: This isn’t a one-time training session. HSBC’s AI is designed to continuously learn and adapt. As new data is fed into the system, the AI refines its “understanding” of financial crime, resulting in increasingly accurate risk assessments over time.


4x More Suspicious Activity Identified: The AI has significantly boosted HSBC’s ability to detect genuine suspicious activity, leading to a potential four-fold increase in identifying financial crime.

Reducing False Positives by 60%: The AI effectively reduces false positives by 60%, freeing up investigators to focus on real threats.

Faster Investigations: HSBC’s AI cuts the data analysis cycle time from a cumbersome 30 days down to a mere 2-3 days, enabling swifter investigations and a quicker response to potential financial crime.

Network Connections Exposed: The AI’s network analysis capabilities help identify suspicious linked accounts, unearthing connections that might be missed in traditional investigations.

The success of HSBC’s AI system highlights the potential role of AI in combatting financial crime. Financial institutions could leverage AI to enhance detection accuracy, reduce investigation times, and protect the integrity of the financial system. As AI technology continues to evolve, we can expect even more sophisticated solutions to emerge in this area.

Predictive Asset MaintenanceAnglian Water

This utility company is piloting an AI-powered predictive maintenance system developed by InfoTiles to proactively manage the health of their biofilters. Biofilters are essential components in wastewater treatment plants, and their proper function is crucial for preventing pollution events.

Here’s how the system works:

Data Acquisition: Data from 76 biofilters across 24 sites is collected through SCADA (Supervisory Control and Data Acquisition) sensor systems. This data includes sensor readings, flow rates, and weather data obtained from publicly available sources. 

Building a Rich Information Picture: The collected data is fed into InfoTiles’ AI model. This model combines the various data streams to create a real-time picture of the operational health of each biofilter.

Predictive Analytics: The AI analyses the data to predict potential issues with biofilter performance. This allows for early intervention before critical failures occur.

The initial phase of the pilot project focused on monitoring the performance of biofilter rotating arms and sludge blanket levels in settlement tanks.


Reduced Risk of Biofilter Failure: The AI system is identifies potential problems with biofilters before they escalate into critical failures, preventing pollution events at Anglian Water’s treatment plants.

Optimised Resource Allocation: By predicting maintenance needs, Anglian Water can schedule maintenance activities more effectively enabling a reduction in the number of emergency callouts, particularly during evenings and weekends.

Data-Driven Decisions: The AI system provides data-driven insights to optimise biofilter operations  allowing informed maintenance decisions based on real-time data analysis.

By leveraging AI and real-time data, Anglian Water can maintain the health of critical infrastructure components, ultimately ensuring a more reliable and sustainable water treatment process. 


So, what makes a strong candidate for AI development?

Generally we would score potential use cases against three criteria:  

1. What is the quantifiable impact this initiative will have in meeting one of our objectives?  

2. How risky is it for the business? Can we mature the solution in a safe way? Is there proven capability in the marketplace we can adopt? 

3. How much will it cost the business? What resources we will have to invest to make this happen?

Setting out a clear decision making framework for your organisation is crucial to ensure that scarce resources are focussed on the projects that can deliver maximum value.


AI’s potential is undeniable, and like any powerful tool, it’s important to use it safely. Instead of a big bang approach, start small and scale smartly. Begin with a pilot project, or test and learn in a controlled environment. This allows for rigorous testing and refinement before a larger rollout, minimising the risk of disruption. 

Additionally, create a supportive incubator environment. This includes early team engagement to explain the solution and benefits, fostering understanding and reducing resistance. Provide your teams with comprehensive training to effectively utilise AI. Identify and support “change champions” who can advocate for the solution, address concerns, and ensure smooth adoption. This comprehensive structure, coupled with ongoing encouragement, is crucial for overcoming the learning curve and paving the way for broader AI adoption within your organisation.


In amongst the hype there are tangible examples of AI being deployed with meaningful benefits. Momentum is clearly building as the enablers we discussed in the second newsletter bed in and companies begin to move beyond experimentation into action.  

The examples highlighted demonstrate the breadth of AI as a field and the optionality that exists in terms of how to create solutions.  

The sheer scale of opportunity and seemingly endless marketplace of AI providers needs to be addressed systematically, as with all other change initiatives, to ensure the solution created stays true to the business needs.  

At Cross 8 we have an extensive track record of leading transformations for clients. Our team have the experience and knowledge needed to help you understand your use case, select the right solution and integrate this new technology to meet your expected outcomes. If you would like to know more contact or