PROCEDURE FOR GENERATING A QUOTATION FOR MOTOR VEHICLE REPAIR

Description

BE2025 / 5004 2 The invention aims to facilitate and expedite the drawing up of a quotation for the repair of damage to motor vehicles. SUMMARY OF THE INVENTION 5 A method for generating a quotation for the repair of damage to a motor vehicle comprising the steps of: training a first AI model with repair data from various motor vehicle repair companies, where the repair data includes historical information regarding, among other things, repair procedures, costs, and time required to repair different types of damage to 10 motor vehicles, preferably at least partly derived from old quotations from the various motor vehicle repair companies; further training the first AI model with repair data from a specific motor vehicle repair company, where the repair data includes historical information regarding, among other things, repair procedures, costs, and time required to repair 15 different types of damage to motor vehicles, preferably in the form of old quotationsprepared by motor vehicle repair company; receiving images of a damaged motor vehicle; characterizing one or more damages to the damaged motor vehicle based on the received images of the damaged motor vehicle using a second AI model trained with data from undamaged motor vehicles, whereby the second AI model undertakes, among other things, the following steps: recognizing the model of the motor vehicle from the images, comparing the images of the damaged motor vehicle with images of undamaged motor vehicles of the same model from the database and identifying and locating damages based on the comparison, identifying and locating various motor vehicle parts, and associating the damages and motor vehicle parts based on the said identifications and locations; determining one or more repair processes for the motor vehicle repair company using the first AI model, whereby the determination was based on thecharacterized one or more damages and repair data originating from the motor vehicle repair company, and where the specified repair process is suitable for repairing the one or more characterized damages; generating one or more quotations for the motor vehicle repair company with the first AI model based on the one or more specified repair processes.35 The described method offers a solution to the time-consuming and error-prone traditional quotation process by automatically generating a quotation for 2025 / 5004 BE2025 / 5004 3 motor vehicle repair companies based on images of a damaged motor vehicle. For this purpose, AI models are used to characterize the damage and predict a suitable repair method, based on which an accurate quotation is subsequently drawn up. 5 DETAILED DESCRIPTION Unless otherwise defined, all terms used in the description of the invention, including technical and scientific terms, have the meaning as theybe generally understood by the specialist in the technical field of the invention.10 For a better assessment of the description of the invention, the following terms are explicitly explained. “A”, “the” and “the” in this document refer to both the singular and the plural unless the context clearly implies otherwise. For example, “a segment”15 means one or more than a segment. The terms “comprise”, “comprising”, “consist of”, “consisting of”, “provided with”, “contain”, “containing”, “encompass”, “encompassing”, “contain”, “containing” are synonyms and are inclusive or open terms that indicate the presence of what follows20, and that do not exclude or prevent the presence of other components, features, elements, members, steps, known from or described in the standard technique. The term “user” refers to the persons who may make use of the current invention. These are primarily employees of motor vehicle repair companies, but may possibly also be external experts or owners of a damaged motor vehicle.The term “motor vehicle” refers to a motor-driven vehicle that is primarily designed for transport on the road, such as cars, trucks, buses and motorcycles, but can also refer to agricultural vehicles, ATVs or campervans. The invention concerns a method according to claims1 for generating a quotation for the repair of damage to a motor vehicle.35 The process of preparing a quotation for the repair of damage to a motor vehicle is traditionally a time-consuming and complex process for 2025 / 5004 BE2025 / 5004 4 motor vehicle repair companies. The quotation procedure requires many labor-intensive resources, such as a thorough physical inspection by an experienced mechanic to determine the damage, the manual documentation of findings, and the calculation of costs based on parts, labor hours and the complexity of the repair. Variations in the extent of damage, motor vehicle models and the required5 repair methods make it difficult and time-consuming to prepare consistent and accurate quotations. This process often requires extensive experience and an accurateto make estimates of both the extent of the damage and the costs and time required for the repair. Small errors in the estimate can lead to financial losses or customer dissatisfaction.10 The described method offers a solution to the time-consuming and error-prone traditional quotation process by automatically generating a quotation for motor vehicle repair companies based on images of a damaged motor vehicle. For this purpose, AI models are used to characterize the damage and predict an appropriate repair method, based on which an accurate quotation is subsequently drawn up. A first AI model has been trained with historical repair data from various motor vehicle repair companies, where the repair data includes information, among other things, about repair procedures, costs, and the time required to repair different types of damage to motor vehicles, preferably in the form of old quotations drawn up by motor vehicle repair companies. A quotation includesdetailed information regarding the individual cost items. For example, a quote for the repair of a broken headlight and a scratch on the front bumper of a car includes information regarding the price for the replacement of the headlight, the costs for painting the bumper, and the number of labor hours required for both repairs. In addition, the quote states which materials were used, such as the paint and the headlight unit. By processing old quotes into the AI ​​model, the model can learn which steps are necessary for different damage cases, how costs are structured, and how much time such a repair typically takes. This information enables the AI ​​model to predict the accurate repair procedure and the corresponding quote in future cases based on characterized damage. 35 However, the repair process is highly dependent on the specific working methods and techniques employed by each motor vehicle repair company. Different companies may use unique techniques, tools and materials, as well as 2025 / 5004 BE2025 / 5004 5variations in the efficiency and experience of their employees. To make an accurate quote estimate specifically tailored to a particular motor vehicle repair company, the first AI model is additionally trained with historical repair data from that specific company. This data includes, among other things, information about previous repairs, costs, materials used, and the duration of 5 different repair methods. By feeding the first AI model with this company-specific data, it learns which processes and costs are characteristic of the motor vehicle repair company in question. This leads to a custom-made quote that takes into account the unique circumstances and practices of the company, which significantly increases the accuracy of the quote and thereby contributes to improved customer satisfaction. To characterize the damage, a second AI model is trained with images of both damaged and undamaged motor vehicles, whereby the second AI models how different damages can visually appear on an image.15Based on this information, the AI ​​model can recognize the type of motor vehicle by comparing images of the damaged motor vehicle with images of other damaged and undamaged vehicles of the same type, which are stored in a database and labeled with the corresponding information regarding the parts present and damage. Through further comparisons with these images, the second AI model can identify the damaged parts and accurately characterize the damage. For example, it recognizes the depth of a scratch or dent based on reflections and contrasts in the image, and detects internal damage based on visible anomalies on the outside of the motor vehicle. Based on the damage characterized by the second AI model, the first AI model can determine automatic and / or more appropriate repair processes and the corresponding quotation, taking into account the specific circumstances of the motor vehicle repair company. In an execution form comprising the said images, close-up photos or videos ofspecific damage, as well as various photos or videos showing a broader view of the motor vehicle, preferably from different angles. In one implementation form, 'template matching' is used for the step of comparing the images of the damaged motor vehicle with images of undamaged motor vehicles of the same model from a database. 2025 / 5004 BE2025 / 5004 6 In one implementation form, the second AI model applies semantic segmentation to determine the exact locations of the damage. This means that the model classifies each pixel in the images as belonging to a specific part and / or damage area. In one implementation form, Convolutional Neural Networks (CNNs) are used for the second AI model. CNNs are very suitable for recognizing visual patterns in complex image data. In one implementation form, Random Forests are used for the first AI model. RandomForest is suitable for processing structured data and recognizing patterns. In one implementation form, SupportVectorMachines (SVMs) is used for the second.AI model. SVMs are suitable for separating data into different classes, and the second AI model is applied to classify different types of damage.15 In one implementation form, Gradient Boosting is used for the first AI model. Gradient Boosting works by iteratively correcting errors from earlier predictions, whereby each subsequent model focuses on the remaining errors of the previous one. This continuous optimization ensures that the AI ​​model becomes increasingly accurate20 as it processes more data, making it highly suitable to fulfill the function of the first AI model. In a preferred implementation form, the first and / or second AI model uses one or more of the following techniques: vision language model; advanced25 computer vision; expert systems (knowledge systems); hybrid approaches. It goes without saying that future techniques can also be implemented. In a preferred implementation form, the first / or second AI model uses one or more of the following architectures for data processing: real-time30processing systems; batch processing systems; cloud-based system; edge computing systems; distributed processing system. It goes without saying that future architectures can also be implemented. In an implementation form, the method comprises the step of displaying information35 on a graphical user interface (GUI), where the information comprises at least one, or a combination of the following elements: the characterized one or more damages, at least one generated quotation, a duration of the specified 2025 / 5004 BE2025 / 5004 7 repair processes and / or a duration of individual sub-processes within the specified repair process. The GUI allows users to easily access relevant data. This makes the system not only more accessible but also more interactive, enabling the repair process to be monitored and controlled more efficiently. Displaying the characterized damage allows users to verify which damage has been detected, which is an important step for the accuracy of thequotation. In a further implementation form, the GUI provides insight into the estimated duration of the repair, both for the entire process and for previously separate sub-processes, enabling motor vehicle repair companies to better manage their expectations and planning. This increases the transparency of the repair process, reduces potential misunderstandings, and offers users the ability to proactively intervene if necessary. In an implementation form, adjustments to a specific repair process and / or generated quotation are displayed on the GUI, and the method comprises the steps of requesting permission for the said adjustments via the graphical user interface, and automatically adjusting the quotation based on the granted or denied permission. 20 By proactively involving the user in the decision-making process regarding possible adjustments to a specific repair process or a generated quote, transparency is increased and adjustments are prevented from being implemented without the user being informed. Requesting permission via the GUIensures direct and efficient communication with the motor vehicle repair company. This makes it possible to make quick decisions without delays, which contributes to a more streamlined and efficient process. When a user agrees to certain modifications or refuses certain modifications, the quote is automatically adjusted based on the granted or refused consent. This ensures that the quote is always up-to-date and fully corresponds to the actual scope of the repair and the user's expectations. In an execution form, the method comprises the steps of repairing the damaged motor vehicle according to a specific repair process, obtaining 35 visual material of the repair, automatically recognizing objects in the visual material and automatically updating the generated quotation based on the recognized objects and / or a time at which the objects are recognized, the recognized 2025 / 5004 BE2025 / 5004 8 objects comprising one of, or a combination of, the following elements:employees, tools, motor vehicles, motor vehicle parts. In a further form of execution, the image material is obtained using one or more cameras installed above the workplaces. In an alternative form of execution, the image material is obtained using portable cameras, worn by the employees of motor vehicle repair companies. Based on this information, it can be determined which repair activities have been completed, allowing the system to dynamically update how much work remains to be done and whether extra costs need to be included. If unexpected complications arise, such as additional labor or parts required, this can be immediately incorporated into the quotation. This leads to a realistic and accurate cost estimate. In a further implementation form, the motor vehicle repair company receives direct updates on the progress of the repair based on the recognized objects. This approach results in a more transparent repair process, which allows the employees of the motor vehicle repair companycan make better-informed decisions and thereby work more efficiently. 20 In another or further form of implementation, the first AI model is used to analyze the image material. Because the first AI model is already trained with various repair processes, it can recognize the activities faster. In addition, the first AI model can immediately apply the obtained information to adjust the generated quotation. 25 In another or further form of implementation, object detection is applied to analyze the activities of employees in the workshop in real-time. The system identifies employees and tracks their movements, providing insight into what work is being performed and whether this is in line with the specified repair process. This information enables the system to further refine the quotation by correctly tracking the progress of the repairs. It can also be recorded which employees perform which tasks, and how quickly they perform them. Based on this, the time required for the various activities is better predicted, wherebya more accurate estimate of the quotation can be made.35 In another or further form of execution, object detection is applied to identify tools and / or parts used for the 2025 / 5004 BE2025 / 5004 9 repair. This increases the accuracy with which the work of the employees can be determined and thereby improves the prediction of the quotation. In one form of execution, after the repair of the damaged motor vehicle, a final calculation is made on the basis of which a definitive quotation is drawn up, whereby the final calculation is based on the objects identified in the obtained image material and the time corresponding to the identified objects, from which the duration of the various repair actions and the means used is determined. During the repair process, the model automatically records the time spent on the vehicle. This system provides a real-time overview of the exact labor time per motor vehicle. This enables the motor vehicle repair company togenerate a detailed and accurate final quotation. In a further or alternative implementation form, the first AI model processes purchase invoices for the damaged motor vehicle, whereby references such as the license plate, file number, and / or chassis number are automatically recognized, and on the basis of which the quotation is generated. This automated approach minimizes administrative errors and increases efficiency, while the company can offer customers a more transparent overview of costs. In another or further implementation form, the method includes the step of displaying information on a graphical user interface (GUI), where the information includes at least one of, or a combination of, the following elements: the progress of the repair in real-time, adjustments in the specific repair process, changes in a generated quotation; and where the information is based on the recognized objects. This allows users to be informed immediately when changes occur in the repair process.what consequences this has for the quotation. This ensures rapid communication within the motor vehicle repair company, which increases the efficiency of the work and reduces the chance of errors. In addition, this offers the possibility to keep the owner of the motor vehicle being repaired directly informed of the progress of the repair. As a result, the owner remains better informed about the repair process, which leads to better customer satisfaction. 35 In another or further form of implementation, the information on the progress of the repair includes photos of the motor vehicle being repaired. This provides visual evidence of the work performed, thereby increasing the transparency and reliability of the process. Photos provide a more complete picture of the progress of the repair so that any errors can be detected more easily and users have a clearer picture of the progress of the repair. 5 In a further form of implementation, the said photos are used to the second AI-to train the model. These images help the model capture variations in damage patterns and recognize subtle deviations, such as hidden internal problems that might otherwise be overlooked. Furthermore, feedback from employees indicating when the model has not correctly identified specific damage can provide valuable information for further optimization of the model. As a result, the AI ​​model remains relevant and can adapt to changing damage repair processes, leading to a more efficient process and increased customer satisfaction. 15 In an implementation form, the method comprises the step of training the first AI model based on the recognized objects and / or the time at which the objects are recognized. The recognized objects in the visual material, such as for example which employees are working on specific tasks, which tools are being used and which materials are involved, offer valuable information for training the first AI model because they provide insight into the operational processes of themotor vehicle repair company. This data enables the model to recognize patterns and evaluate the progress of various work methods, leading to more accurate cost and time estimates specifically tailored to the motor vehicle repair company. Furthermore, the model can identify which 25 combinations of workers, tools, and materials are most efficient, allowing it to optimize the performance of the repair process. This enables the AI ​​model not only to generate more accurate quotes but also to provide recommendations for the most efficient repair methods. 30 In an implementation form, the method comprises the step of providing an upload function integrated into a graphical user interface (GUI), where the upload function is configured for uploading data relating to one or more damages, and where during the step of receiving data from a damaged motor vehicle the said data are received35 via the said upload function, and where the graphical user interface (GUI)is suitable for implementation on a mobile device, preferably a mobile phone and / or tablet. This enables users to document the damage at the location of the 2025 / 5004 BE2025 / 5004 11 motor vehicle, allowing a quotation to be generated immediately at the location of the damaged motor vehicle. Employees of the motor vehicle repair company and the owner of the damaged motor vehicle can thereby upload a photo of the damage directly while inspecting the damage and obtain an accurate quotation. As a result, both parties immediately have a realistic idea of ​​the situation, which speeds up the entire repair process. In one form of implementation, the method comprises the step of receiving a description of the one or more damages, and where the step of characterizing the one or more damages is performed based on the aforementioned description using a Natural Language Processing (NLP) model. Basing the characterization of damage on a description improves accuracy,because these explanations may contain details about aspects that are not visually visible, such as hidden structural damage or other internal problems. Through the use of Natural Language Processing (NLP), the description can be automatically and accurately analyzed so that relevant information is extracted. This step enables a more complete understanding of the damage and a more accurate characterization of the damage, which makes the generated quotations more accurate. In one implementation form, the second AI model and Natural Language Processing (NLP) are combined into a single Visual Language model. This increases efficiency because image and language are processed simultaneously, reduces the complexity of the architecture, and improves the interpretation of multimodal relationships, whereby more accurate and contextually relevant results are achieved. 25 In an implementation form, the first AI model and the second AI model are integrated into a single AI model. Combining the first and second AI model into a single model ensures that the information from both processes is shared and accessible.within the same system. This allows the AI ​​model, when generating quotes, to take into account not only the complete information of the detected damage, but also the specific repair processes and costs directly. This leads to a more complete and accurate analysis. Because the data from both processes are integrated, the system can better determine which repairs are most suitable for the identified damage, resulting in more accurate quotes. 35 In an implementation form, the first AI model and / or the second AI model are designed as composite systems comprising multiple sub-AI models, each tailored to perform a specific function within the overall framework of the AI ​​model. These sub-AI models can operate independently or in a coordinated manner to achieve the desired functionality. In the first AI model, individual sub-AI models can, for example, specialize in tasks such as analysis of historical data, cost forecasting, and pattern recognition, whereby each model focuseson a separate aspect of the repair estimation process. Similarly, the second AI model can contain sub-AI models dedicated to image segmentation, object detection, and feature classification, which work together to comprehensively analyze and characterize vehicle damage. In certain configurations, these sub-AI models can function as separate entities, communicating via well-defined interfaces to exchange data and insights. Alternatively, they can be integrated into a uniform architecture, utilizing shared resources and interconnected layers to perform all tasks cohesively as a single AI model. In an implementation form, the method comprises the step of comparing the generated quotation, or a part thereof, with a quotation, or a part thereof, drawn up by the motor vehicle repair company, where the compared quotations, or parts thereof, relate to similar repair processes or similar sub-processes within repair processes, and the adaptation of the first AI model.and / or approving or modifying the generated quote, or part thereof, on the basis of the comparison. Through this comparison, discrepancies between the quote generated by the AI ​​and the quote from the repair company can be identified and corrected. This ensures that the AI ​​quote test becomes increasingly better aligned with the actual repair processes and associated costs. Furthermore, any unrealistic estimates of the quote are prevented by this. Furthermore, it promotes a continuous learning process for the first AI model. The quotes prepared by the motor vehicle repair company reflect the actual repair process and actual costs. By comparing the generated quotes with quotes prepared by the motor vehicle repair company, the model can adjust and optimize its algorithms based on the observed discrepancies, so that the generated quotes correspond more accurately with the actual costs of the repair. In an implementation form, the method comprises the step of adapting a specificrepair process based on information provided by one or more employees of the 35 motor vehicle repair company, where the information relates to a change in the specified repair process and / or a change in the characterized one or more damages. This promotes the accuracy of the specified repair process, 2025 / 5004 BE2025 / 5004 13 because employees can share valuable insights and practical experiences that influence the execution of the work. The employees have valuable experience enabling them to recognize deviations between the characterized damage or the specified repair process and reality. This information is therefore essential to estimate a realistic quotation. In addition, it increases the flexibility5 of the repair process. When employees report changes, such as the discovery of hidden damage during the repair, the first AI model can immediately update the quotation so that all users are aware of this new information. 10 In a further form of implementation, the method comprises the steps of repairingthe damaged motor vehicle according to a specific repair process, and the training of the first AI model based on information provided by one or more employees of the motor vehicle repair companies, where the provided information relates to the repair. The experience of the employees of the motor vehicle repair company is valuable information because it concerns specific damage and therefore cannot be obtained anywhere else. This information may concern specific actions that must be performed and were not yet foreseen by the first AI model, or incomplete characteristics of the damage by the second AI model. This may also concern photos of newly encountered damage or the progress of the repair. By using this information to train the first AI model, the accuracy of the generated quotation can be significantly increased because the repair process can be mapped out more accurately by the first AI model. brought. In addition, this information provides valuable insights into the working methods of thevarious employees of the motor vehicle repair company, whereby the quotation can be more accurately tailored to working methods specific to the motor vehicle repair company. In what follows, the invention is described by means of non-limiting examples that illustrate the invention, and which are not intended or should not be interpreted to limit the scope of the invention. EXAMPLES EXAMPLE135 The first AI model is trained by making use of historical repair data originating from various motor vehicle repair companies. This includes 2025 / 5004 BE2025 / 5004 14 different types of data, such as old quotations, detailed cost overviews for parts and labor, the duration of various repairs, and information on used materials. Possible sources of this information are internal databases of the repair companies, such as software for damage management and invoicing, in which previous repairs and associated costs are recorded. In addition, data canderived from documentation, such as work orders and reports, drawn up by employees after completing repairs. Data from customer feedback and service history, or market prices for parts and labor can also be used to train the first AI model, to ensure that the AI ​​model gets a more complete picture of the variables that influence repair costs and time.10 By combining these diverse information sources, the first AI model can identify patterns and correlations that help accurately predict costs and time for future repairs based on the damage to the motor vehicle. The training of the first AI model begins with collecting and preparing15 this data, correcting incomplete or inconsistent information. Subsequently, machine learning algorithms, such as Random Forests or Gradient Boosting, are applied to the dataset to identify patterns and correlations. During the training process, the model learns, for example, which recovery methods are typical.are applied for specific damage cases and how they are related to the associated costs and time. After the training, the AI ​​model is validated, whereby it is tested with a separate dataset, possibly consisting of 10% to 20% of the collected data, but this can also be between 5% and 10%, or between 20% and 40%. This dataset contains examples of previous repairs for which the actual costs are already known. The model predicts the costs for these repairs based on the input variables it has learned during the training. Subsequently, these predictions are compared with the actual costs to determine how accurate the model is. This process uses various statistical metrics, such as the mean absolute error (MAE) and the mean percentage error (MAPE), to get a clear picture of the model's performance. The results of the validation are carefully analyzed. If the accuracy of the predictions does not meet if it meets expectations, adjustments are made to the AI ​​model. This canentailing that hyperparameters are tuned, such as the number of trees in a Random Forest or the depth of the trees, to improve the generalization capacity of the model. Additionally, new data, such as more recent repair information, can be added to enrich the training of the model and further increase its ability to make accurate predictions of the repair process and its costs. 2025 / 5004 BE2025 / 5004 15 The second AI model is trained using a dataset consisting of images of both damaged and undamaged motor vehicles. In the first step, these images are collected from various sources, including internal databases of motor vehicle repair companies, in which previous damage cases are recorded, and other databases containing reference photos of different types of motor vehicles. In this process, each image is labeled with relevant information about the damage, such as the type of damage, the location, and its severity. Subsequently, the collected images are cleaned and prepared for analysis. This entails that the images arenormalized and annotated, so that the second AI model can learn which visual features correspond to which specific damage. The annotations contain important information, such as identifying parts of the vehicle and classifying the types of damage, such as dents, scratches, or breaks. The training of the second AI model takes place using techniques such as Convolutional Neural Networks (CNNs), which are particularly effective for image recognition. During this phase, the model learns by analyzing the annotated images and identifying patterns and features specific to different types of damage. The AI ​​model uses a process of feedforward-back propagation, comparing its predictions with the actual labels and adjusting its parameters to improve accuracy. After the initial training, the model undergoes a validation phase, using a separate dataset of images to test the model's performance. The accuracy of the predictions is evaluated using statisticalmetrics, such as accuracy and the F1 score, are used to determine how well the model performs in correctly identifying and classifying damage. Based on the validation results, adjustments are made to the model during the optimization phase, such as refining hyperparameters or applying data augmentation techniques to increase the variability of the training data. This ensures that the model becomes more robust and is better able to generalize to new, unseen images. Finally, a Natural Language Processing (NLP) model is trained with a dataset consisting of textual descriptions of damage, derived from employees of motor vehicle repair companies or from customer feedback, in which customers share their experiences and comments regarding the damage and repairs. Additionally, information and reports from previous work orders can be used, in which detailed descriptions of previous damage cases are recorded. Data from 2025 / 5004 BE2025 / 5004 16 external platforms or databases, such as insurance claims in which damage isdescribed, are included. Furthermore, manuals or documentation of the vehicles themselves can be analyzed to integrate technical terms and definitions, which helps the model understand specific jargon. Finally, online forums and discussion groups where damage repair is discussed can provide valuable textual information about many different damage cases. In the first step, these descriptions are collected and annotated, recording important elements such as the nature of the damage, the severity, and specific circumstances of the accident. These annotations can also contain contextual information, such as previous repair experiences with similar damage. Next, the textual data is cleaned and prepared for analysis. This entails removing irrelevant information and stop words, and normalizing the text, for example by using techniques such as tokenization and lemmatization. These steps ensure that the model can focus on the essential content of the descriptions.The training of the NLP model may be performed using techniques such as word embeddings (e.g., Word2Vec or GloVe) and recurrent neural networks (RNNs) or transformers. During this phase, the model learns to recognize patterns and mantic relationships in the text, enabling it to understand contextual information and extract relevant aspects of damage descriptions. The model applies a process of feedforward and back propagation, in which it compares its predictions with the actual annotations and adjusts its parameters to improve accuracy. After the initial training, the model is validated with a separate set of textual data to test performance. The accuracy of the predictions is evaluated using statistical metrics such as precision, recall, and F1 score, to assess how well the model is able to extract relevant information from the descriptions. Based on the validation results, optimizations are performed, such as adjusting hyperparameters and refining the training dataset toto increase variability. This leads to a more robust model that is better able to provide accurate characterizations of damage based on the textual input. The first AI model is being retrained to specifically tune into a motor vehicle repair company by using historical data that is representative of that company's unique working methods. For this purpose, old 35 quotations, detailed work orders, and specific cost overviews characterizing the company are collected. During the retraining, the model optimizes its parameters and algorithms based on this specific data, so that it learns from the work processes, 2025 / 5004 BE2025 / 5004 17 cost structures and patterns specific to the company. This new information includes variables such as parts and labor prices, which can vary significantly depending on the company and region. Additionally, employee working methods can vary; for example, some companies prefer replacing parts rather than repairing them, while others 5employ repair techniques. The presence of advanced equipment is also important, as companies with more modern technology can organize their processes more efficiently. The use of specific materials, such as OEM parts versus aftermarket options, as well as the time employees require to perform certain actions, are crucial for the training of the model. Furthermore, differences in staffing capacity, customer segments, and warranty and service conditions can influence the company's operational methods. By integrating all these diverse and specific data, AI can model the unique characteristics and operational needs of the repair company, enabling it to generate more accurate quotes that are tailored to the reality of their daily activities. To effectively characterize a new damage case, image information of the damage is used. The image information can include photos and / or videos that provide a clear representation of the external damage to the motor vehicle, such as dents in thebodywork, scratches on the paintwork, or broken lights. These visual details are essential for identifying the extent of the damage and determining the necessary repairs. A description can also be added: "The dent in the left door was caused by a collision with another car, where the impact also resulted in a tear in the rubber seal." This description captures not only the visible damage but also details regarding the cause and possible internal damage that is not visible in the images. The images and description of the damage can be easily uploaded via a graphical user interface (GUI) implemented on a mobile device such as a smartphone or tablet. Employees or owners of damaged motor vehicles can take photos of the damage to the vehicle with their mobile phone or tablet and send them directly via the upload function in the GUI. The interface offers user-friendly buttons to select or take photos, and a text box in which a description of the damage can be entered, for example detailsabout35how damage occurred or any internal problems that are not visible in the photos. The GUI is designed to work quickly and efficiently on mobile devices, so that users can document damage on-site themselves and share this information in real-time 2025 / 5004 BE2025 / 5004 18 with the motor vehicle repair company. This allows the damage repair process to be started faster. The second AI model analyzes the received images of a damage case in a structured process to accurately characterize the damage. First, the second AI model optimizes the image for further analysis. This may involve, for example, normalizing the image, where brightness and contrast are adjusted to improve the visibility of the damage. Subsequently, the AI ​​model performs object detection, identifying specific parts of the vehicle, such as doors, bumpers, and windows. This can be achieved using techniques such as Convolutional Neural Networks (CNNs), which are trained to recognize the contours and features of various vehicle parts.Alternative methods such as YOLO (You Only Look Once) and SSD (SingleShotMultibox Detector) can also be used to detect objects quickly and efficiently. For example, the model can detect and mark a door in an image to determine where damage has occurred. Subsequently, the AI ​​model evaluates the specific damage to the identified parts by capturing various aspects of the damage. This includes analyzing the intensity, color, and texture of the pixels in relevant areas, with which the model can determine whether the damage is superficial, deep, or structural. For example, the model can determine that the right door has a 5 cm wide dent, but also that the rim is damaged, which may indicate a need for both dent removal and repainting. Additionally, the model can also identify internal damage, such as torn wiring or damaged lighting systems, by looking for unusual deviations or changes in the areas surrounding the damage. The first AI model then proposes a repair process based on the characterized damage.to analyze the second AI model and link it to the historical repair data of the specific motor vehicle repair company. Assume that the damage consists of a crack in the windshield, deformation of the window frame, and a slight displacement of the window seal. Based on the damage classification, the model recognizes that the windshield must be replaced, the frame must be aligned, and the seal possibly needs to be reapplied. Based on this detailed evaluation, the first AI model selects repair procedures that best fit the available resources and working methods35 of the motor vehicle repair company. For example, if the company specializes in glass repairs using advanced equipment for the precise alignment of window frames, the model takes this into account and selects a procedure in which 2025 / 5004 BE2025 / 5004 19 the frame is aligned directly in the workshop. In addition, the model calculates the time required for each of these steps based on the work experience of theemployees and previous repairs that the company has carried out. Ultimately, the AI ​​model establishes a repair process with a clear step-by-step approach, including estimated time and costs, resulting in an efficient and personalized repair plan for the specific company. In addition to an image of the damage, a description may also be added. This description is analyzed by the NLP model by first tokenizing the text, whereby the words are divided into separate units such as damage categories and vehicle parts. The model then recognizes relevant terms and patterns, such as "the dent in the front bumper" or "scratches on the right door," by using trained word vectors that capture the semantic relationship between words. The NLP model can also identify synonyms and technical jargon, such as "crack" or "tear," and thereby classify the extent and severity of the damage. This analysis refines the overall characteristics of the damage. After the second AI model and the NLP model have characterized the damage, canThe first AI model then determines one or more repair processes based on this. There may be different methods suitable for repairing the damage that differ from one another. For example, if there is a damaged wheel, the AI ​​model can suggest replacing the wheel completely as one option, or repairing the wheel using specialized equipment as an alternative option. Each process is tailored to the capabilities and preferences of the company, whereby the costs and time for the repair are carefully calculated. The quotes for each process are automatically generated by the AI ​​model, taking into account the costs of parts, such as the wheel or any mounting materials, and the labor costs required to carry out the specific repair process. For the replacement option, for example, the quote may turn out higher due to the price of a new wheel, while the repair option entails lower costs but may require more labor.The model can also take into account factors such as the use of specialized equipment or the hiring of additional personnel. All this information is displayed on a graphical user interface (GUI), which enables employees of the motor vehicle repair company to easily view the specified repair process and any alternative repair processes, along with the associated quotations. The GUI presents a repair process clearly, with details such as the materials required, the time the repair will take, and the estimated total costs. Employees can browse through any alternative repair processes and compare options. The GUI also offers the possibility to make adjustments to the proposed repair processes or quotations. For example, if an employee decides that a wheel repair is feasible without completely replacing the wheel, they can implement this change via the GUI. Or the price of necessary materials, available materials, or available equipment may have changed without the initialThe AI ​​model is aware of this. It may also occur that the employee detects hidden damage during the repair, requiring the repair process to be adjusted. All these changes can be adjusted by the employees via the GUI, and the first AI model subsequently automatically adjusts the quotation and the specific repair process based on this new information. The first AI model compares generated quotes with quotes prepared by the motor vehicle repair company, whereby the AI ​​model recognizes any discrepancies. In doing so, parts of quotes are compared that relate to comparable damage, actions, or materials. This means that the AI ​​model analyzes specific elements, such as the cost of parts or the required labor time, in the context of previous repairs that correspond to the current damage. When a clear discrepancy is established, such as a significant deviation in the estimated costs or duration, the first AI model attempts to determine the cause of this deviation and judge whether an error has been made.If, for example, the deviation is greater than a certain percentage or absolute difference established based on historical data, such as 20% higher costs for labor or parts compared to previous repairs of similar damage, the model can flag this deviation as a potential error. Additionally, contextual information can be used, such as the nature of the damage and the associated repair methods. For example, if it encounters a situation where a simple scratch is flagged with unrealistically high repair costs, the AI ​​model can conclude that a likely error has been made. By applying such criteria, the AI ​​model can effectively determine which deviations need to be investigated further and which should be corrected as an error. When an error is made, the generated quote is automatically adjusted. If, however, in the event of a significant deviation it is not immediately clear whether an error35 has been made, this discrepancy is displayed on the graphical user interface(GUI), so that an employee of the motor vehicle repair company can check it. 2025 / 5004 BE2025 / 5004 21 During the repair of the damaged motor vehicle, the specified repair process and the associated quotation can be automatically adjusted based on footage obtained in real-time from the workshop of the motor vehicle repair company. This footage can be captured with cameras5 installed above the workstations or with wearable cameras, such as bodycams worn by employees. These cameras continuously record the progress of the repair and send the images to the AI ​​model for analysis. An AI model, trained with an extensive dataset of labeled images of various repair scenarios, analyzes the acquired images to determine the progress of the repair. The model can, for example, identify object detection tools used by workers, such as screwdrivers, dent removal tools, or spray guns, and can also recognize vehicle parts such asdoors, bumpers, or wheels. By detecting which tools are being used and which part of the vehicle is being worked on, the AI ​​model can determine which specific repair activities are being performed, such as removing a damaged door or straightening a bumper. Additionally, by using pose estimation, the AI ​​model can track and recognize the actions of workers, such as removing parts, reassembling parts, or preparing a vehicle for paint spraying. If, for example, the model detects that a bumper has been replaced instead of repaired, the quote can be automatically adjusted to include the extra costs for the new part. Furthermore, the AI ​​model can also recognize specific employees. The quotation can then be further adjusted by taking into account the specific pace at which individual employees work. The model can, for example, detect that an experienced employee works faster than a less experienced colleague, which influencesthe expected time for completing certain repairs. By linking the recognized employee to historical data regarding their performance, the AI ​​model can make an accurate estimate of the time and costs required for the repair.30 If an employee who typically takes longer for certain actions is recognized, the quote can be automatically adjusted to reflect extra time and labor costs. Based on the information extracted from the imagery through object detection, the first AI model can be further trained to be even more accurately tailored to the specific motor vehicle repair company. The first AI model thereby learns how different objects and actions influence the repair processes and can use this data to make future quotes 2025 / 5004 BE2025 / 5004 22 more accurate. For example, if the AI ​​model recognizes that a specific employee uses a replacement part instead of a repair method, it can learn that this method typically results in higher material costs and fewer labor hours.entails. Additionally, by analyzing how different employees perform specific repairs, the model can learn which actions require more time or cost, depending on the skills and experience of the employees involved. By adding this information to the existing database of the first AI model, time-cost estimates can be further improved. Furthermore, the first AI model can be trained with data on unexpected complications emerging from the imagery, such as internal damage that only becomes visible during the repair. The model can then learn to anticipate such situations in future forecasts. This continuous feedback loop, in which the model continuously collects new data and learns from the images of the repair process, ensures that the first AI model becomes increasingly accurately tuned to the unique processes, personnel, and preferences of the specific motor vehicle repair company. As a result, it can generate increasingly realistic quotes that not only take standard into account.repair processes, but also with the specific characteristics and dynamics of the company. All information regarding the progress of the repair, obtained from the visual material of the work, is displayed on a graphical user interface (GUI) so that employees of the motor vehicle repair company can easily stay informed of the status of the repair. This information may include, among other things, which specific repairs are currently being performed, the time spent on each task, and which parts have been used or replaced. For example, the GUI may indicate that the right door is currently being dented and that the process is expected to take another 30 minutes, while also providing a warning if additional damage is detected. This transparency promotes better cooperation among employees, because they have direct insight into each other's work and can react faster if problems arise or if there is miscommunication. This can improve overall repair efficiency, which leads to faster completion of the work and highercustomer satisfaction. 35 During the repair, various photos of the motor vehicle can be taken during the process to document the progress of the work. These photos can be taken at key moments, such as before 2025 / 5004 BE2025 / 5004 23 the repair begins, during key steps such as the removal of damaged parts, and after the completion of the repairs. This visual documentation provides a clear overview of the progress and can reveal any complications or changes in the damage. The photos taken can easily be uploaded via the graphical user interface (GUI). This allows employees of the repair company to immediately share the images with external experts, such as insurance advisors or claims adjusters, or with the owner of the motor vehicle themselves. This makes it possible to maintain transparent communication regarding the status of the repair and any necessary adjustments resulting from new findings.For example, if traditional damage is discovered during the repair, photos can be uploaded to visually substantiate the situation, ensuring that both the client and external parties remain well informed about the progress and any changes in the repair process. The first AI model can use these photos to improve the accuracy of the quote. By analyzing the images, the AI ​​model can gather additional data on the repairs performed and the materials used, and integrate this information into its estimates. This ensures that the quote is better aligned with the reality of the repair. 20 In addition, these photos can also be used to further train the second AI model and / or the first AI model. By labeling and annotating the images with information about the repairs performed and the associated damage, the AI ​​models can characterize damage even more accurately and determine a repair process, because they continuously learn to recognize more unique damage cases and unique operational characteristics of the motor vehicle repair business. EXAMPLE 2The present invention relates to a method for accurately estimating the costs of repairing a damaged vehicle and generating a detailed invoice. The method begins by retrieving current damage information via direct communication with external systems via Application Programming Interface (API) endpoints. The API query system is co