DigitalTransformationTalk: Industry AI – What you need to know

On 1 October 2024, Digital Transformation host Kevin Crane was joined by Dave DuVarney, Principal, Baker Tilly; Mark Brewer, VP Service Industries, IFS; Mike Gosling, IT Service Platforms Manager, Cubic Transportation Systems; and Peter Pearce, Principal, Baker Tilly.

Views on news 

AI increases operational efficiency for manufacturers by minimizing, or completely removing, repetitive tasks. It also offers manufacturers greater agility in how they face industry changes and business requests by learning and understanding information that humans could overlook, while AI-generated insights empower manufacturing leaders to make informed, data-driven decisions. Use cases where manufacturing can leverage AI include content summarisation for engineers when sending an assessment to clients after a visit; recommendation engines to support cross-selling and upselling; predictive maintenance through anomaly detection; planning and schedule optimisation; forecasting and simulation for demand forecast; running what-if scenarios; and training workforce to operate machines.

How to identify and bridge skill gaps

Route scheduling tools are becoming increasingly sophisticated, as they look at a diverse set of factors such as the service-level-agreement (SLA) of the exercise, the tools and skills required, the location, the certification and access points and they will allocate and schedule the right engineers automatically on the basis of these data sets – while also constantly learning from the planning processes. The upside is that when there is an incident that requires human intervention, the automated system will keep on doing its scheduling uninterrupted. What ChatGPT can add to this is an interface on top of other applications, where users can interact with digital tools in natural language. In the long term, this can enable LLMs to talk to other systems via APIs. ChatGPT can also be used in tandem with OCR technology to ingest data that is less structured – this can save thousands of hours for companies that must deal with tens of thousands of documents per month. 

Airplane engine manufacturers, for example, use ML to gather flight data from their airline customers and leverage that to extend the maintenance interval. In call centres, the technology can be used to supervise agents, where a dashboard will indicate if there is some problem on a call that requires intervention. PSO can improve SLAs up to 30% in the matter of months on a contract where several engineers have 3-4 jobs a day on the customer’s system, thus enabling a company to grow without extra skills or resources. LLMs can also help detect micro events, which are undetectable otherwise but can lead to problems. There are certain components that you must have in place when starting AI initiatives. The first is data governance, then there is core infrastructure and the knowledge of how to use it, data privacy (many vendors are addressing now the issue by preventing sensitive data getting out) and driving adoption, but it’s also key that you have an innovation engine within the organisation. Once you have a use case, adopt an agile approach starting with a proof of concept and a pilot and see if it gets a bit of traction. If it does, it can be fed back into the development mechanism to continue building on it. It’s very useful to conduct a business value analysis to see in what order you should execute the different use cases – quantifying them will also help with selling them to the board. 

The panel’s advice

• Some organisations have embedded innovation teams across every function, while others have a centralised team. The key is that business stakeholders are tightly involved in the innovation process.
• Never try and eat the AI elephant in one bite. Build small-scale solutions first on top of the AI that you already have.
• AI is not about doing the same thing better or cheaper – it’s about doing things in a completely different way.