General Business Processes and AI performance

AI Impact in Specific Sector´s Processes

Source: Perform AI Erasmus+ Project

Developing a specific AI tool for a particular task within a particular sector of activity, the AI methodology that supports it (ML, DL, RPA, etc.) will be configured to achieve a relevant result, often different between sectors. An RPA-based solution for livestock farming will allow vaccination of the animals that need it on the correct date. In contrast, an RPA solution for a third-party accounting services company will enable it to convert an invoice received on paper into data that its accounting system can directly process. The methodology used, RPA, is the same, but the solution developed and the result sought are different in each sector. In this sub-module, we will analyse the sectors of activity shown in the graph above. Let´s see.

Insurance

In the Insurance Sector, the main AI applications are data submission automation, Risk assessment and related competitive price-fixing.


Workflow	Tasks	AI Methodologies	AI TOOL Process Application Description	Outcomes Description
Submission	Submission Processing	RPA (Robot Process Automation) / Optical Character Recognition (OCR) / Natural Language Processing (NPL)	They extract information from unstructured data through optical character recognition (OCR) and natural language processing (NLP). These technologies can eliminate the need to manually review each document from traditional or non-traditional sources and help underwriters capture and classify helpful information	AI can significantly improve submission processing and save valuable time for more productive tasks.
Risk Assessment	Client Profile Analysis	Big Data / Machine Learning (ML) / Predictive Analysis	Using machine learning models and other analytical techniques, underwriters can deepen their understanding of the risk associated with a client's profile. Using data from internal and external sources, these models can learn from the past and predict the risk profile of new submissions.	Underwriters can save a significant amount of time allocated to data analysis and make more informed decisions since these models are better than humans at recognizing patterns in extensive data.
Insurance Pricing	Profile Risk and Profitable Price Fitting	Big Data / Machine Learning (ML) / Predictive Analysis	By using machine learning models, insurance companies can price risk more competitively. Moreover, AI-driven systems can enable real-time data-based dynamic pricing.	Having a better understanding of risk enables underwriters to decide on profitable and fair pricing, that is, the risk profiles. These can increase profitability through improvement in workflows, and pricing strategy, reduce time-to-market and ensure customer satisfaction and retention.

Source: Diagnostic Tool - LiveCircular Canvas Project.

Logistics & Transportation

In the Logistics and Transport Sector, the myriad of movements, transactions, and intermediate stock points in the journey of products from their origin to their final destination generate a multitude of opportunities for the application of AI.

Key applications include:

- The management of material movements and the resources involved.
- The control of capacities and means of transport used.
- The organisation and optimisation of distribution and delivery routes.
- Planning and forecasting of demand and associated revenues.


Workflow	Tasks	AI Methodologies	AI TOOL Process Application Description	Outcomes Description
Logistic Administration	Shipping and Goods Movements Data	RPA (Robot Process Automation)	Predictive analytics estimates customer demand by extracting and organizing historical data by applying the analytical test to them.	With various demand forecasting methodologies, the organization can improve their decision-making processes in Capacity planning.
Logistic Administration	Warehouse Management	Machine Learning (ML) / Sensors / RPA (Robot Process Automation) / Autonomous Vehicles	Goods are received at the warehouse (Inbound), identified, sorted, processed, packaged, and pulled for shipment(outbound); all steps are automatically carried out with minimal negligible flaws and even tracked the same and eliminated in subsequent stages.	Combining Machine Learning with IoT sensors and real-time monitoring data creates an intelligent warehouse that can also be linked to the centralized data processing unit, so that volume of order processing increases overall productivity: in pick-and-pack processes, slotting routes for packages, parking, and delivering the right package.
Logistic Resources Maintenance	Predictive Maintenance	Data Mining / Machine Learning (ML) / RPA (Robot Process Automation)	AI-driven systems automatically gather and analyze data about assets and warn about possible breakdowns. Such systems can accumulate information about failures, draw up statistics and schedule repairs based on this statistical data.	If a provider can guarantee the employment of available and reliable assets, it will help gain a competitive edge and the trust of service consumers.
Supply Chain	Logistic Planning	Predictive Analytics / Machine Learning (ML)	Predictive analytics is estimating customer demand by extracting and organizing historical data by applying the analytical test to them.	With various demand forecasting methodologies, the organization can improve their decision-making processes in Resource planning, Risk assessment, sales, and marketing operations Strategy, etc.,
Routing	Route Design & Optimization	Data Mining / Machine Learning (ML) / RPA (Robot Process Automation)	Route optimization techniques use shortest-path algorithms to identify the most efficient route for logistics trucks to deliver the product.	With route optimization, AI helps to keep track of road traffic, reduce fuel consumption, improve air quality and urban planning deliveries, identify the driver's free time and facilitate parking. It helps speed up the shipping process, considerably reducing shipping costs, and ultimately improves customer satisfaction.
Yield Strategy	Yield Optimization	Data Mining / Machine Learning (ML) / RPA (Robot Process Automation)	AI systems can make the entire process faster, safer and smarter. This helps facilitate timely delivery to the customer as per the commitment. Automated systems accelerate traditional warehouse procedures, thus removing operational bottlenecks along the value chain with minimal effort to achieve delivery targets.	With Yield optimization, you can determine which inventory areas are performing well and which are not. You will understand the specific elements or circumstances of these areas, exploiting them to maximize efficiency and boost your revenue.
Logistic Demand Forecasting	Demand Prediction	Data Mining / Machine Learning (ML) / Predictive Analysis	You can use AI and Machine Learning to predict demand or improve demand forecasting and warehouse management.	Based on experience, you will get a detailed analysis of all the factors influencing the demand and find valuable insights.

Source: Diagnostic Tool - LiveCircular Canvas Project.

Education

From the Educational Centre Administrative and Academic management to personalized learning content, tools, feedback and assessment methods, AI is increasingly present in the Education sector.


Workflow	Tasks	AI Methodologies	AI TOOL Process Application Description	Outcomes Description
Education Centre Management	Education Centre Administrative Tasks	RPA (Robot Process Automation)	Systems for the academic management of courses, teachers and students, and administrative and documentary management associated with the educational centre.	Saving efforts in academic and documentary management of the centre. Accuracy, compliance control, standardization and consistency of data. A multitude of pre-designed reports can be executed effortlessly.
Learning	Personalized Learning Contents Definition	Machine Learning (ML)	The itinerary and contents that best suit the learner's style are extracted and proposed by analyzing the learner's learning profile and the learning database of a multitude of similar profiles. This proposal is continuously adjusted according to the results of its use by the student and the evolution of their learning style.	Personalized electronic learning content customized to the learning styles and preferences of the pupil.
Learning	Accesibility & Universalization	Machine Learning (ML) / RPA (Robot Process Automation)	Creation of profiles and personalized curricula suggestions, updated as the student covers stages and consolidates learning outcomes. The system can follow the student to any place and time where it is necessary.	Universal access to global classrooms. Life-long and life-wide learning supportive systems.
Learning	Learning Assistance	Voice & Visual Assistants / Virtual Environments	Virtual human guides, virtual reality games, and facilitators for use in various educational environments. An AI-powered virtual teaching assistant is a program designed to improve a student's learning experience. Many online college courses require students to learn something by themselves.	AI-driven virtual teaching assistants are designed to take some load off the faculty's shoulders. With machine learning capabilities, these programs don't just provide answers to common questions; they can analyze the learning process and make suggestions to optimize the approach to education.
Learning	Smart Conten	Virtual Platforms / Smart Content Generators	AI uses algorithms to personalize the student experience. It learns your way of learning. At the same time, it generates data to analyze the needs of individual students and the classroom as a whole. This technology also attempts to condense textbooks into valuable tools for exam preparation, such as true or false questions.	The expected result is a greater return in student learning, achieved through the personalization of content and the learning and assessment autonomy provided by the system. Effectiveness: greater availability and greater personalization. Efficiency: less personal effort on the part of the teacher.
Feedback Management	Instant Feedback	RPA (Robot Process Automation) / Machine Learning (ML)	This tool enables students to receive formative feedback on their academic skills based on criteria set by teachers. Students can iterate on their assignments before they hand in a final version.	Learner autonomy implies potential acceleration in the learning process. Standardization of the feedback received by all students, which is agreed upon in advance by a cadre of teachers.

Source: Diagnostic Tool - LiveCircular Canvas Project.

Retail

Ai is present in the Retail sector with customer shopping assistance, virtual trial rooms, assortment and stock level definition, automated payment methods and competitive pricing.

Workflow	Tasks	AI Methodologies	AI TOOL Process Application Description	Outcomes Description
Retail Customer Experience	Purchase Assistance	Chatbots / Recommendation Systems / Visual Search Systems / Voice Search Systems	A chatbot is a computer program that communicates through audio or text. Chatbot uses advanced Natural Language Processing techniques, but simple systems scan for keywords within the input and pick a response with the most analogous or similar word patterns from databases.	Such programs are designed to adequately imitate how a human could behave as a communicating partner. Provide a higher level of customer service, send notifications about new collections, and suggest similar products. Chatbots target customer needs and solves their problems efficiently and transparently.
Retail Customer Experience	Virtual Trial Rooms	Augmented Reality / Virtual Reality	They work by having the customers' image in the stores, and they can select the apparel which needs to be tried on, making one have a realistic picture of the outfit without even literally trying it. This makes the entire process really convenient and effective without too much in-store mess.	Help customers save time and money from travel expenses and the convenience of their home; the customer finds the perfect outfits with all elements perfectly matched in just minutes. AI scanners can receive enormous sales within the same customer base by giving them a real-time picture of what would suit them.
Retail Demand Forecasting	Store Demand Planning	Big Data / Machine Learning (ML) / Predictive Analysis	Data scientists could open a new world of possibilities to business owners by extracting anomalies and correlations from hundreds of Artificial Intelligence and Machine Learning models to plan the store demand level for a period attending to many factors.	AI in the retail supply chain can be used for restocking - calculating the demand for a particular product by considering a history of sales, location, weather, trends, promotions, and other parameters. It resulted in studied cases, with up to a 30% reduction of in-store shelf gaps.
Merchandising Management	Spaces Distribution and Assortment on the shop shelves/websites	Big Data / Machine Learning (ML) / Predictive Analysis / Computer Vision / Web Analytics / IoT	Sales data analysis by shop and shelf location can be stored and modelled to improve performance by addressing seasonal and other influencing factors.	Improved sales results per shop visit and optimization of the facing assigned to each item on the shelves.
Store Inventory Management	Inventory Needs Calculation	Big Data / Machine Learning (ML) / Predictive Analysis / Robot Process Automation (RPA)	AI-based technology combined with other algorithms of machine learning can help stores regulate and keep a check on product categorization. It also would help get the real-time inventory to the store to bring in the required stocks each day to avoid missing out on a customer.	Increases customer satisfaction helping the customer to easy in-store products found. This can help retail stores to be able to get all the products piled up and ready for sale in their inventory timely. Reduces Inventory process costs.
Retail Cash Management	Cashier Free Systems	RPA (Robot Process Automation) / IoT	Shopping technology works when you pull something out from the shelf or put it back when you walk out of the shop; the store account will deduct your purchase within a fraction of a second. In other cases, machines scan the barcodes of the purchased goods and are also ready to accept digital modes of payment.	By cutting down employee compensation, increasing efficiency, and reducing wait time, the robotization of stores had significant savings on operational expenses.
Retail Price Strategy	Price Simulation/Sensitivity	Big Data / Machine Learning (ML) / Predictive Analysis	AI technology for retail stores visualizes the likely outcomes of multiple pricing strategies. It reads data within seconds from the business's earlier transactions and the market for the product's actual cost, promotional activities, and sales figures in the market and keeps its par for generating profits within the same threshold with higher efficiency.	AI technology for retail stores helps to set prices for their products more competitively to thrive in the market. AI reads and interprets the statistics according to demand, seasonal trends, characteristics, customer choices, the release date of new models of the same items, etc.

Source: Diagnostic Tool - LiveCircular Canvas Project.

Livestock

Individualisation in animal feeding, health control, reproduction management and animal output management, are the main contributions of AI in Livestock sector.


Workflow	Tasks	AI Methodologies	AI TOOL Process Application Description	Outcomes Description
Livestock Management	Balanced Feeding / Intake Composition	Machine Learning (ML) / Sensors / Drones	A livestock management system powered by computer vision that lets farmers monitor cows. The camera feed is analyzed in real-time, and results are forwarded to farmers’ mobile phones.	Smart cameras monitor animals' physical activity, feeding patterns, and overall aggressiveness to inform farmers about their herd's well-being and ensure that the appropriate amount of feed is dispensed, increasing herds' productivity.
Livestock Management	Animal Disease Monitoring	Facial Recognition Systems / Movement and activities recognition	A chip monitors vital information about the animal, including temperature variations, feeding patterns and reproductive stages through the smart sensor, and can detect diseases 48 hours before they manifest, allowing for timely medical attention. The information for each animal is synched and stored in a cloud platform.	The main goals are to utilize the data to maximize production and limit stress levels on the cows. Boosts farming profit margins by an average of 15%. Detects health issues with up to 95% accuracy
Livestock Management	Vaccine Administration	Robotic System	Robotic injection system to deliver vaccines and reproductive medicines to domestic animals on the dairy farm.	The robotic injection system reads the RFID tags attached to the cow's ear and gets health-related information and vaccination record. If the cow requires an injection, it is directed towards the injection site and the injection mechanism position itself to deliver the medication to the cow's neck. Reduction of costs and 100% accuracy.
Livestock Management	Estrus Management	Big Data / Sensors / Machine Learning (ML)	The cow's neck collar (with motion sensors) collects all types of data 24 hours daily. AI components of the dairy automation system process the collected data to provide insights on the heat stress, change in feeding efficiency and the estrus of the cow.	The occurrence of the estrus cycle results in the release of special hormones that affect the cow's behaviour and movement. The AI component compares the recently collected data (about movements) with stored data and can predict the ovulation period of the cow in advance. Since the ovulation period starts after 24 to 32 hours of the onset of "Standing heat," the farmer has enough time to prepare for artificial insemination of the cow in heat.
Output Management	Antibiotic Usage	Big Data / Sensors / Machine Learning (ML)	The solution allows animals to be monitored 24/7. If the system detects patterns of behaviour that are consistent with illness—for example, pigs drinking more than usual or moving abnormally—it can immediately alert the farmers, so they can take targeted action quickly and avoid the need for drugs.	Detecting the early signs of sickness as well as the quick recovery of cows is key for animal health, welfare and a reduction of antibiotics. A study has shown that with Ida, an intelligent dairy assistant, the recovery time of cows can be reduced by more than half.
Output Management	Emotional Contagion Prevention	Machine Learning (ML)	ML-based AI can help us identify the variables of emotional contagion based on vocalizations, olfactory cues, etc., to detect the outburst of a specific disease or stress.	This has the potential to help us improve the living conditions and well-being of farm animals significantly.

Source: Diagnostic Tool - LiveCircular Canvas Project.

Agriculture

Vi kan finde AI-løsninger spredt ud over hele landbrugsprocessen fra jordforberedelse, vanding, såning, kontrol med afgrødernes sundhed, høst osv..


Arbejdsgang	Opgaver	AI-metodologier	AI TOOL Proces Beskrivelse af ansøgningen	Resultater Beskrivelse
Jordovervågning	Kontrol af jordbundssundhed	Computer Vision / Deep Learning / Droner	Værktøjet identificerer potentielle fejl og næringsstofmangler i jorden ved hjælp af billeder fra mobiltelefoner. Analysen foretages af softwarealgoritmer, som korrelerer bestemte løvmønstre med visse jordfejl, planteskadedyr og sygdomme.	Kontrol af jordkvalitet og dens egnethed i hver enkelt del af landbrugsprocessen. Reduktion af indsatsen og besparelser i doseringen af de løsninger, der er nødvendige for at opnå den rette jordkvalitet.
Overvågning af afgrøder	Kontrol af afgrødesundhed	Sundhedskontrol af afgrøder Computer Vision / Deep Learning / Droner	Udnyttelse af computervision og deep-learning-algoritmer til at behandle data, der er optaget af droner og/eller softwarebaseret teknologi til overvågning af afgrødernes sundhed	Bekæmpelse af ukrudt ved hjælp af præcisionssprøjtning.
Forvaltning af afgrøder	Vanding	Fjernbetjeningssensorer og Arduino-teknologi	Automatiske plantevandingsanlæg er plantet på marken via trådløs teknologi til drypvanding.	Denne metode sikrer jordens frugtbarhed og en effektiv udnyttelse af vandressourcerne.
Høst af afgrøder	Høstopgaver	Landbrugsrobotter	Autonome robotter klarer vigtige landbrugsopgaver som f.eks. høstning af afgrøder i større omfang og hurtigere end menneskelige arbejdere.	Reducer omkostningerne ved høstopgaverne og erstatte den mangel på arbejdskraft, der reducerer sektorens vækst.
Forvaltning af høst	Høst af prædiktive analyser	Maskinlæring	Maskinlæringsalgoritmer i forbindelse med satellitter til at forudsige vejret, analysere afgrødernes bæredygtighed og evaluere gårde for forekomst af sygdomme og skadedyr.	Modeller til sporing og forudsigelse af forskellige miljøpåvirkninger af afgrødeudbyttet. Datakilderne omfatter temperatur, nedbør, vindhastighed og solstråling samt sammenligninger med historiske værdier.

Kilde: Diagnostisk værktøj - LiveCircular Canvas Project

Services

Similar to product manufacturing, AI can help the services sector in the service research and conceptualization process. Specific contributions in the Service sector are related to the Service Production process, with a strong focus on:

- Service resource planning and allocation to assure service experience to the customer.
- Service Quality Control on time, since QC in services should be mainly preventive because the production and consumption time is simultaneous, there is no stock like in products to amend any quality problem before consumption.


Workflow	Tasks	AI Methodologies	AI TOOL Process Application Description	Outcomes Description
Service Design	Service Research, Conceptualization, Structuration & Testing	Big Data / Data Mining / Machine Learning (ML) / Natural Language Processing / Web scraping / Predictive Analysis / RPA (Robot Process Automation) / Internet of Things (IoT).	The AI technology collects the data from a selected target audience to define and test the main service concepts, structure the basic, complementary and governance services, and finally, tests de servuction and the satisfaction associated.	Market research needs to be done ASAP to maintain relevance. Delay in delivery often results in outdated insights and inaccurate sentiment analysis. AI market research tools, Increase efficacy and reduce process costs.
Service Operation	Service Resource Planning	Data Mining / Machine Learning (ML) / Predictive Analysis	Based on historical resource usage and demand forecasts, taking into account several variables, the algorithm will generate demand levels for each service, identify bottlenecks and recommend resource allocations.	Adjust Servuction and Resource Availability and Usage (People, Spaces and Tools) to Forecast Demand. Service Operation costs reduction.
Service Operation	Service Quality Control (QC)	Data Mining / Machine Learning (ML) / Predictive Analysis / RPA (Robot Process Automation) / Internet of Things (IoT)	Use of smart cameras and related AI-enabled software for service inspection tasks. Including facilities, customer participation, service team, service tools and other service environment elements.	Improved quality inspection at speeds, latency, and costs beyond the capabilities of human inspectors

Source: Diagnostic Tool - LiveCircular Canvas Project.

Consumer Packaged Goods

I CPG-sektoren kan vi ud over generelle AI-løsninger til processer finde andre specifikke løsninger, der er rettet mod:

- Produktdesign, test og forbedring
- Ressourceallokering i fremstillingsindustrien
- Kvalitetskontrol af de indgående råmaterialer og komponenter og produktionsresultater.


Arbejdsgang	Opgaver	AI-metodologier	AI TOOL Proces Beskrivelse af ansøgningen	Resultater Beskrivelse
Produktdesign	Produktforskning og -udvikling	Big Data / Maskinlæring (ML) / Behandling af naturligt sprog / Webscraping / RPA (Robot Process Automation)	AI-baserede markedsundersøgelser leverer resultater i næsten realtid. AI-teknologien indsamler data fra en udvalgt målgruppe og overvåger og scanner automatisk nøgleord eller emner.	Der skal foretages markedsundersøgelser hurtigst muligt for at bevare relevansen. Forsinket levering resulterer ofte i forældede indsigter og unøjagtige analyser af stemninger. AI-markedsundersøgelsesværktøjer, øger effektiviteten og reducerer procesomkostningerne.
Produktion	Teknisk kontor	Datamining / maskinlæring (ML) / forudsigelsesanalyse / RPA (robotprocesautomatisering)	Identificere potentielle forbedringer, der er afledt af mindre ændringer af tekniske produktspecifikationer.	Reducerer produktomkostningerne og/eller øger kundetilfredsheden.
Produktion	Indkøb (råmaterialer og komponenter)	Datamining / maskinlæring (ML) / forudsigelsesanalyse / RPA (robotprocesautomatisering)	Ved at forudsige efterspørgslen kan RPA-løsninger forudse de forskellige leverandørers komponentbehov, og de nødvendige ordrer automatiseres under hensyntagen til de minimumsordrer, der er forhandlet med leverandørerne.	Det reducerer omkostningerne ved komponent- og råvareopfyldningsprocessen og justerer omkostningerne ved sådanne lagre.
Produktion	INPUT/OUTPUT Logistik	Datamining / maskinlæring (ML) / forudsigelsesanalyse / RPA (robotprocesautomatisering)	For at forudsige lagermangel og håndtere den ved at arrangere nye forsendelser kan CPG-virksomhederne overvåge produktplacering og lagerbeholdning ved hjælp af IoT-enheder og andre kilder som transaktionsdata og endda tredjepartskilder som f.eks. vejret.	AI-værktøjer kan fjerne manuel indsats og øge effektiviteten i logistikken.
Produktion	Oplagring	Datamining / maskinlæring (ML) / forudsigelsesanalyse / RPA (robotprocesautomatisering)	Produktionsprognosen giver os mulighed for at opstille indkøbsprognosen. Vi kan bestemme, hvilken blanding af komponenter der skal lagres i hver periode, og hvor hurtigt de vil blive omdannet til færdigvarelager.	Justering af de nødvendige investeringer i faciliteter og andre menneskelige og tekniske ressourcer, der er nødvendige for at forvalte de indgående og udgående lagre i perioden.
Produktion	Produktion	Datamining / maskinlæring (ML) / forudsigelsesanalyse	AI giver os mulighed for at indarbejde de mange forskellige datakilder, der er nødvendige for at producere resultater, der er præcise nok til den moderne virksomhed. AI-baserede systemer kan mere præcist forudsige det nødvendige antal produkter og den frist, inden for hvilken de skal genopføres.	Reducerer overlagring af komponenter og færdige produkter.
Produktion	Produktkvalitetskontrol (QC)	Datamining / Maskinlæring (ML) / Prædiktiv analyse / RPA (Robot Process Automation) / Internet of Things (IoT)	Brug af intelligente kameraer og tilhørende AI-aktiveret software til inspektionsopgaver i produktionen.	Forbedret kvalitetsinspektion med hastigheder, latenstid og omkostninger, der ligger ud over menneskelige inspektørers muligheder.

Kilde: Diagnostisk værktøj - LiveCircular Canvas Project.

Further resources: Videos and/or useful links

PERFORM AI Movies.

PERFORM AI Animations.

References

Business Process Applications. In Depth Guide to the Top 100+ AI Use Cases & Applications. https://research.aimultiple.com/ai-usecases/

AI in Consumer Packaged Goods (CPG). https://www.dataiku.com/stories/reshaping-cpg-industry-with-ai/

AI in Services. Springer Nature Switzerland AG 2019 77 P. P. Maglio et al. (eds.), Handbook of Service Science, Volume II, Service Science: Research and Innovations in the Service Economy, https://doi.org/10.1007/978-3-319-98512-1_5