Example Use Cases

Artificial Intelligence (AI) Maturity Tool

1. Strategy and Management

Feedback mining Feedback mining
The tremendous amount of information collected on public internet, through customer services, surveys and social media is only valuable if you can transpose it into suitable decisions. Through an AI-based analytics and ontology management tool covering 100+ languages. Supporting Sales & Marketing functions, this tool is also used to value production logbooks and operators feedback at a leading process industry player.
Forecast of markets Forecast of markets
To increase the success rate of product launches by using predictive analytics, Natural Language Processing (NLP) and computer vision to capture trends in real-time from a diverse set of publicly available on-line sources (blogs, social media, menus…).

Algorithms specifically adapted to food vocabulary and language agnostic (trained on several Asian languages).
Customisation of new products Customisation of new products
The solution predicts flavours and customer preferences at the pre-production stage, saving money for future food and beverage technology companies that will avoid spending money on low-demand production. AI understands the human perception of taste and preferences, dividing users into different demographic groups and modelling their preference behaviour or predicting what they want before they do it.

2. Products and Services

Optimised product Optimised product
The world's first AI-generated whisky. Predictive models of the distillery are fed by existing recipes (including those of award-winning blends), sales data and customer preferences, and generate new recipes with predicted success and quality.
Virtual engine (digital twin) Virtual engine (digital twin) extends the business opportunities (models)
Sensor engine monitoring for aircraft enables rental models instead of just selling the engines. The current state of the engine can be determined by evaluating the data. Maintenance and servicing can be carried out as needed on this data pool. In addition, performance and consumption data from various flight channels can be evaluated and distributed to the aviation companies.
New product recipes New product recipes
Generating pizza recipes using deep learning technologies by combining two recurrent neural networks (RNN). AI has learned to find non-obvious links between the ingredients of pizza, to understand how to combine the ingredients and how the presence of each influences the combinations of the others.

3. Competence and Cooperation

Problem solving (based on history) Problem solving (based on history)
"Industrial performance is, throughout the world, based on daily rituals enabling priority sharing, problem solving, objective management and team spirit. To help teams solve their problems, AI detects whether a similar problem has occurred in the past. An intelligent solution to capitalize and enhance the experience on the shop floor and not reinvent the wheel anymore."
Shopfloor dashboard Shopfloor dashboard
Many agri-food processes require high frequency data processing to assure shopfloor visibility on the Edge. An IoT Platform offers huge Edge processing capabilities for real-time equipment and quality monitoring and sensitive shopfloor data capturing and usage. Bridging Edge and Cloud apps enables to make the first step towards intelligent and autonomous factories.
Interoperability of information systems Interoperability of information systems
Aggregation and reconciliation of the main information systems (IS) of a production company in order to provide accurate Key Performance Indicators with different levels of time granularity is a difficult endeavour that must be maintained throughout the life cycle of the IS. Easy integration with industrial standards to access trusted data sources and provide the right information at the right time and the right place is therefore always a challenge but has a high benefit.

4. Processes

Optimisation of production process Optimisation of production process
Liberty Aluminium Dunkirk optimises its aluminium production process by electrolysis. The AI system deployed analyses about forty production parameters to recommend the 3 actions that best increase the density of the anodes, an essential parameter to guarantee the performance of the electrolyser.
Failure prediction Failure prediction
Implementation of the software program, for the monitoring and diagnosis of nuclear fuel cell fabrication lines, in order to predict failures and trigger work requests in Computerized Maintenance Management System.
Quality control Quality control
Leveraging the real time collection of dedicated process parameters, AI pattern analysis was developed to identify deviations and non conformities from standard patterns. Early detection is valued by the production team to anticipate non conformities, to improve interactions between operators and process systems, and to stop production before producing non conformant parts.
Classification with machine vision Classification with machine vision
Improve the quality of processes through intelligent optical inspection, combining both a large number of images from the previous vision system, all labelled "good" or "bad", and techniques to increase its image base to increase the number of images with defects (very few defects in microelectronics) and therefore to generate a strong AI.
Waste (object) identification by machine vision Waste (object) identification by machine vision
Industrial recycling is key to make Industry compatible with recent decisions regarding material flows. Many sites process dangerous objects and need intelligent systems to better identify and clean appropriately items. Based on advanced visual and edge inspection capabilities, these solutions help workers on their daily activities.

5. Data

Validation of electronics Validation of electronics
Implementation of AI approaches to automate the analysis of data from its pre-silicon validation processes. Execution tests conducted during the pre-silicon validation phase generate TB of data; the AI enables automatic mining of this mass of data to assist validation engineers in their task of identifying optimal parameter sets.

5. Technology

Design optimisation Design optimisation
With different software, engineers and designers specify the design objectives as well as the different parameters (mass, volume ...) and constraints (materials, process limitations ...). All these specifications are then digested by a Deep Learning algorithm that generates design variants that it considers optimal from the point of view of some constraints on which the engineers/designers can then work.

After reading these AI application examples it is possible to proceed with free AI maturity tool available from the button below.

Start AI maturity assessment