Digital transformation through smart AI and IoT

    Techniche CTO Tom Caldwell recently presented at forecourttech’s virtual event about how AI, machine learning and IoT can help achieve digital transformation. Fellow speakers included the International Forecourt Standards Forum (IFSF) and NACS Europe, while delegates were fuel and convenience retailers from around the world. In this blog, Tom outlines some of the key points from his conference session.

    One of the most pressing challenges which AI can help fuel retailers solve is improving customer service. Automating manual tasks so staff can focus on improving, not doing, or using new technologies to spot certain situations and behaviours which need attention, creates new levels of efficiency and service delivery.

    Download Tom Caldwell’s presentation, ‘Digital transformation through smart IoT and AIOps’.

    Experimental AI in Silicon Valley

    This is where AIOps (Artificial Intelligence Operations) comes in, which is all about applying new, off-the-shelf AI and machine learning capabilities to improve the customer experience.

    During the last year, we’ve been testing new AI image recognition, IoT and machine learning technologies by running a pilot in Silicon Valley across a network of fuel retail sites. The focus has been on using image recognition to automate processes and solve safety concerns by recognising unsafe behaviours and alerting an attendant to take appropriate action.

    AI image recognition models have been trained to recognise certain objects or conditions.

    Being able to identify objects and then connect objects together means different behaviours are recognised. This leads to ‘cognitive AI’, where we identify good behaviours versus bad behaviours or ‘normal’ activity compared to anomalous. For example, identifying that people are smoking, driving away with a pump nozzle still attached to the vehicle, not paying attention while refueling, or spotting incidents such as spilt fuel, fire, or smoke.

    As part of the pilot, AI has been introduced to identify security issues on the forecourt, such as recognising suspicious vehicle activity in and around the forecourt.

    The same techniques are also being used to improve the customer experience inside the convenience store, identifying unwanted situations such as a dirty coffee machine or an unstocked fridge. Or in terms of security, identifying suspicious individuals inside the store who may be about to commit crime.

    At our pilot sites, managers receive an automated daily report of what needs attention, along with a list of network-connected assets that are being monitored for their availability, performance levels and any cybersecurity concerns which may arise.

    These test projects are already delivering immediate value and are only really scratching the surface – the future possibilities are endless.

    Building blocks towards successful AI

    But ahead of any widespread adoption of these new exciting developments, organisations need to understand what is involved in this AI journey and overcome some common challenges.

    First you need to understand your assets. Some are smart and connected to your network, some are legacy and unconnected, but many are critical, revenue-generating assets for your business. Data on all assets needs to be collated into an asset inventory or register, with the correct data architecture, in order to make the first steps towards AI possible.

    Experience also shows there is a convergence of interests and governance when it comes to adopting AI and IoT, which brings together IT, FM and other teams that may not have worked so closely before. New conversations are required to determine who owns what and where responsibilities lie. Often these departments are very siloed and not accustomed to working together, but all sides will have to find a compromise for a project to be successful.

    Take the example of implementing smart IoT devices and assets, or using CCTV cameras for AI image recognition projects. Every smart device or camera has a back-end cloud which has APIs and data, most likely managed by IT. Every smart device widens the threat landscape so security operations need to be involved too. And anything on the forecourt has safety concerns and is related to facilities management in some way.

    Mobilising your AI and IoT strategy

    As alternative fuels and new models of mobility begin to disrupt the industry, some fuel retailers are already on the path to realising their vision of how AI and IoT can benefit their business. And through pilots and experimenting with new technologies, we can learn where they can deliver the most business benefits across the forecourt, store or car wash.

    In summary, any successful AI and IoT strategy should consider the following points:

    • The foundation of any strategy is based on knowing what critical assets you have
    • Ensure you can gather the right operational data that can be collated, analysed and reported on in the cloud
    • Focus on use cases and solving problems that will have a high impact on your future business – it’s not just about the ROI or cutting costs
    • There are a lot of easy-to-use AI and machine learning packages available – you don’t need an army of data scientists to create a successful project – a small team is fine

    UT blog CTA BOFU mobile

    UT data scientist Blog CTA BOFU@2x