The inception of industry 4.0 and smart, autonomous and data driven processes will bring significant opportunity for the agri-food sector to leverage value-based decisions with regards to productivity, waste, risk and efficiency. AI and ML capabilities form part of the greater autonomy that industry 4.0 affords manufacturing sectors, providing the necessary knowledge (from data) to enable humans to make business efficient decisions for them and the wider supply chain. AI has the potential to bring positive social impacts. A Fronter AI report suggested that agriculture was least exposed to labour market disruption because of AI, suggesting that AI will positively enhance the work lives of people in agri-food. Identified as a high potential growth sector, there are endless opportunities for AI application within agri-food; from helping drive prevision agriculture to predict disease outbreaks, weather and yield, livestock monitoring and anomaly detection and supply chain optimisation through predictive algorithms for managing waste, inventory and demand and detection of food safety outbreaks. This project proposes working with 2 large, digitally mature organisations involved in the manufacture and retailing of food in the UK, to develop bespoke generative AI chatbots, housed within the existing Foods Connected (FC) demonstration platform. Housing within a demonstration platform and using representative data will highlight use cases of AI in a safe and secure environment, paving the way and de-risking future AI projects with live data. Both these businesses have used the platform for a number of years to collect various data points, e.g. audits, supplier management, technical due diligence and quality/safety checks in their supply chain, and have a wealth of data of which dummy data can be modelled off. Generative AI chatbots can analyse large amounts of data significantly faster and more precisely than humans, improving data driven decisions and automate administrative tasks and reducing resource waste. From a range of inputs, chatbots using pattern recognition can make predictions that inform users that make key business decisions. Combined with representative data, external sources including food authenticity-based ML models for origin discrimination and alerts or horizon scanning services, could all act as inputs to the chatbot. Users could ask the chatbot, 'How many deliveries of soya in the last month have been sourced from Brazil?'. Outputs or responses could help users determine sourcing strategy risk both from an economical and ethical (social) perspective and adjust practices accordingly in a significantly more efficient way than done so currently.

"Food fraud costs British families £1.17billion per year according to National Food Crime Unit estimates. Due to the nature of adulteration, there are also increased food safety risks associated with consuming adulterated food and drink. Company revenue and brand reputations have been destroyed after implication in food fraud and safety scandals. The latest BRC global standards are pushing retailers to do more on authenticity/integrity testing. Conventional authentication methods are complex, laboratory based, require skilled operators are expensive and lack corroborative methodology. They are failing to meet the needs of the food industry stakeholders. Cibus Analytical are looking to introduce an innovative step change in food fraud detection to address these issues, we are working on a two-tier approach: * Customers will use handheld Near Infrared spectrometers to scan samples at point of sampling and will have access to cloud based chemometric models, that distinguish between authentic and adulterated commodities. Results will be delivered to their location in seconds allowing them to make informed decisions in a timely manner. * Suspicious samples will undergo analysis at our laboratory using our spectroscopy and chemometric model based accredited methods. This will provide confidence and support to our customers in their decision making and will take a fraction of the time of conventional methods. Within this project our aim is to develop and validate our first product offering, making our research more commercially relevant by creating a minimum viable product (MVP) to demonstrate our capabilities to customers and investors. Specifically we will develop and validate our collection of chemometric models for food fraud detection and transfer these to a cloud based environment; develop the agnostic user interface that will automate data transfer, analysis and presentation of results for the field deployable device; validate the applicability of our product offering through a feedback loop of customer trials; develop and accredit the additional laboratory based chemometric models needed to corroborate in-field results."