OLD INTERNAL-ESTIMA AI

[Audio] Our company has been working on an AI-driven estimation platform called ESTIMA AI, which uses machine learning algorithms to estimate project costs and timelines. The platform is designed to provide accurate and efficient estimations for businesses, allowing them to make informed decisions about their projects and operations. One of the key features of ESTIMA AI is its ability to analyze large amounts of data from various sources, including financial reports, construction schedules, and other relevant information. This allows us to identify trends and patterns that may not be immediately apparent to human analysts, enabling more precise estimates. We have developed a robust framework for integrating multiple data sources, which enables seamless communication between different systems and applications. This framework also facilitates the sharing of knowledge and expertise across teams and departments within an organization. By leveraging this framework, our clients are able to streamline their processes, reduce errors, and improve overall efficiency. In addition to its analytical capabilities, ESTIMA AI also offers advanced visualization tools, which enable users to easily interpret complex data and gain insights into project performance. These tools include dashboards, charts, and graphs that provide a clear and concise representation of project metrics, making it easier for stakeholders to understand and communicate project status. Furthermore, our platform is designed to be highly customizable, allowing organizations to tailor the system to their specific needs and requirements. This flexibility enables our clients to adapt to changing market conditions, new technologies, and evolving business strategies. Overall, our vision is to create a comprehensive platform that integrates all aspects of project management, providing a single source of truth for project-related data and analytics. With ESTIMA AI, we aim to empower businesses to make data-driven decisions, drive innovation, and stay ahead of the competition..

[Audio] The ESTIMA AI platform is designed to drive estimation processes with artificial intelligence. This enables accurate and efficient estimations across various projects. By leveraging machine learning algorithms, the platform can analyze large datasets and provide precise estimates. The AI-driven approach ensures consistency and reliability in estimation results. Furthermore, the platform offers real-time insights and updates, allowing users to track progress and make informed decisions. With its advanced features and capabilities, the ESTIMA AI platform is poised to revolutionize the way estimations are made..

[Audio] The project development roadmap outlines key milestones achieved between September 2023 and March 2025. In the first phase, business foundation was established through discussions with the Tride team regarding existing processes and scope. Initially, a comprehensive dataset covering three aircraft types was shared in December 2023. This dataset served as the basis for the first proof-of-concept results, which were reviewed and refined by the Tride team in January-March 2024. An expanded dataset covering five aircraft types was then shared in April 2024, allowing for further evaluation of the model's efficiency. Finally, in May 2024, the validation phase commenced, involving the review and validation of the results against executed packages, followed by additional feedback from the Tride team. Throughout this period, the project progressed steadily, laying the groundwork for future developments..

[Audio] The project underwent extensive testing over a six-month period. This resulted in 26 bug fixes and 35 product enhancements. These enhancements were based on real-time usage and have significantly improved the platform's functionality. During this testing phase, a total of 810 man-hours were spent ensuring the accuracy and reliability of the system. Additionally, 55 packages were tested across various platforms. This rigorous testing process has ensured that the final product meets high standards of performance and stability..

[Audio] The comprehensive technical evaluation of the Estima AI model involved assessing its technical requirements and feasibility. This assessment included evaluating data preprocessing, model training frameworks, algorithms, and deployment strategies. Multiple teams from different departments worked together to conduct this evaluation. They set up the necessary infrastructure, including servers and hosting environments on cloud platforms, primarily utilizing Microsoft Azure. Azure Blob Storage was used for storing raw historical data, while Azure Cosmos DB served as a globally distributed NoSQL database. Additionally, Azure ML Platform hosted the machine learning pipelines for data preprocessing, model training, and inference. Furthermore, Azure Container Apps and Azure API Gateway were established. To optimize the system architecture, modifications were made to support the AI model's performance and scalability. Supervised learning models, specifically Random Forest and Gradient Boosting, were employed for classification and regression tasks, predicting Man-Hours, Spare Parts Usage, and Probability of Findings. These models were selected due to their robustness, interpretability, and capacity to handle high-dimensional data..

[Audio] The frontend modifications include redesigning the user interface to make it more intuitive and user-friendly. The navigation has also been improved, with a focus on layout and responsiveness. User feedback has been incorporated to streamline workflows and enhance accessibility. Several key enhancements have resulted from accuracy improvement discussions. Text normalization has been applied to clean free-text fields by applying techniques such as lowercasing, removing special characters, and trimming extra spaces. Semantic clustering and feature engineering have been implemented to utilize techniques like TF-IDF vectorization and cosine similarity to convert textual descriptions into numerical representations. Supervised learning models such as random forest and gradient boosting have been used for classification and regression tasks. These models were chosen because they are robust, interpretable, and can handle high-dimensional data. Advancements in these areas enable better generalization and reduced noise in the data. Furthermore, DBSCAN clustering has been utilized to group related discrepancies and defects. Moreover, the models have been trained on large datasets to improve performance. Additionally, the incorporation of user feedback has led to enhanced accessibility. Finally, the implementation of semantic clustering and feature engineering has enabled better conversion of textual descriptions into numerical representations..

[Audio] The manual testing process validated the RFQ prediction model's performance against its functional requirements. This validation ensured that the model accurately predicted man-hours and spare part costs. However, the testing also revealed discrepancies, edge cases, and deviations in model predictions that needed to be addressed. To address these issues, a comparison was made between predicted and actual values using the Compare Estimate API. This comparison helped to refine the model by identifying areas where it required improvement. The testing feedback was thoroughly documented and used to enhance the model's logic and robustness. Additionally, vulnerability assessments and penetration testing were conducted to identify potential security risks associated with the AI system and hosting environment. These assessments revealed critical vulnerabilities that needed to be addressed. Based on the findings, security patches and best practices were implemented to protect sensitive data and maintain system integrity. Furthermore, a comprehensive review of all project documentation was conducted to ensure accuracy, completeness, and alignment with project objectives and stakeholder expectations. The final documentation was prepared for knowledge transfer, audit readiness, and future reference..

[Audio] The results from our material analysis show that the actual costs are consistently higher than estimated costs using Estima AI. Manual estimates are 85% accurate, while Estima AI predictions are 90% accurate. Actual cost checking reveals that Estima AI's predicted costs are 90% accurate, while manual checking shows an accuracy rate of 90%. The End Of Life analysis indicates that Estima AI's predicted costs are 90% accurate, whereas manual analysis shows an accuracy rate of 90%. However, there may be some discrepancies in certain specific EOL checks, such as VT-ATQ and VT-IZU, where the accuracy rates range from 65% to 70% due to issues with task data and input file inaccuracies..

[Audio] Estima AI's predictions were found to be highly accurate, with 90% of actual unbillable hours predicted correctly. This accuracy was also reflected in its predictions for End-of-Life (EOL) manual hours, where 95% of actual values were matched. However, the manual effort required for checking the Code of Conduct (C) was significantly higher than expected, with only 80% of actual values matched. These findings suggest that Estima AI can effectively streamline estimation processes and reduce manual effort..

[Audio] The Estima AI system has been developed to provide accurate and efficient solutions for estimating costs and analyzing data. The system uses advanced algorithms to analyze historical data and forecast future trends. This enables the estimation of costs associated with projects and products, as well as the identification of potential risks and opportunities. The system can be used to generate reports and make recommendations based on the analyzed data. The accuracy of the estimates is improved through the use of machine learning techniques and artificial intelligence. The system can also be integrated with other software systems to provide a comprehensive view of the project's financial situation. The integration allows for real-time monitoring and control of the project's budget and resources. The system provides a detailed breakdown of the estimated costs, making it easier to identify areas where cost savings can be achieved. The system can also be used to analyze and optimize supply chain operations. By using the Estima AI system, organizations can improve their ability to estimate costs and make informed decisions about investments and resource allocation. The system can help reduce costs and increase efficiency by automating routine tasks and providing insights into areas where costs can be optimized. The system can also be used to predict future costs and revenues. The accuracy of the estimates is further improved through the use of advanced statistical models and machine learning algorithms. The system can be customized to meet the specific needs of each organization. The customization allows for the creation of tailored reports and recommendations that are specific to the organization's goals and objectives. The system can also be integrated with other business systems to provide a seamless experience for users. The integration allows for real-time collaboration and communication among team members and stakeholders. The system provides a user-friendly interface that makes it easy for users to navigate and understand the output. The system can also be used to track and monitor progress towards goals and objectives. By using the Estima AI system, organizations can improve their ability to estimate costs and make informed decisions about investments and resource allocation. The system can help reduce costs and increase efficiency by automating routine tasks and providing insights into areas where costs can be optimized. The system can also be used to predict future costs and revenues. The accuracy of the estimates is further improved through the use of advanced statistical models and machine learning algorithms..

[Audio] The proposed efforts post-implementation of AI feature difference in manual hours average for calculating unbillable per C check work scope are 216 MH, while the manual effort quantification as per existing process yields 27 MH. This represents a reduction of 189 MH in manual effort required. Similarly, the average manual hours for calculating unbillable per HMV (6Y/12Y/18Y) checks decreased from 144 MH to 24 MH, resulting in a reduction of 120 MH in manual effort. Additionally, the average manual hours for calculating unbillable per EOL checks decreased from 112 MH to 28 MH, resulting in a reduction of 84 MH in manual effort. These reductions indicate significant improvements in efficiency and productivity through the implementation of AI-driven estimation..

[Audio] ## Step 1: Identify the main components of the text The text appears to be comparing two types of manuals, RFQ (Request for Quote) manual and Estima AI Area of Impact Manual, with regards to effort quantification. ## Step 2: Extract key information from each component RFQ manual: - Average MH for calculating Unbillable per C = 12 MH - Average MH for calculating Unbillable per HMV (6Y/12Y/18Y) = 16 MH Estima AI Area of Impact Manual: - Average MH for calculating Unbillable per C = 2 MH - Average MH for calculating Unbillable per HMV (6Y/12Y/18Y) = 4 MH ## Step 3: Compare the average MH values for each type of manual Comparing the average MH values, we can see that the RFQ manual has higher values than the Estima AI Area of Impact Manual. ## Step 4: Calculate the difference in MH values between the two manuals To calculate the difference, subtract the average MH value of the Estima AI Area of Impact Manual from the average MH value of the RFQ manual. For C: 12 MH - 2 MH = 10 MH For HMV (6Y/12Y/18Y): 16 MH - 4 MH = 12 MH ## Step 5: Determine the proposed efforts post-implementation of AI feature Proposed Efforts post implementation of AI Feature are not explicitly stated in the provided text. ## Step 6: Summarize the findings The RFQ manual has higher average MH values than the Estima AI Area of Impact Manual for both C and HMV checks. The difference in MH values between the two manuals is 10 MH for C checks and 12 MH for HMV checks. The final answer is: There is no numerical answer to this problem..

[Audio] The company has implemented several key features to enhance its functionality. The Estimate Navigator dashboard provides a clear view of the estimated costs and timelines associated with each project. This feature aligns with the report format generated by the platform, ensuring seamless integration and streamlined workflow. The screen preload feature allows users to efficiently predict material consumption and analyze usage patterns. The incorporation of Service Bulletins, Customer Work Orders, and in-house cards enhances the accuracy of estimates, including turnaround times. Access controls have been established to restrict viewing and interaction to relevant data only. The API integrations on the GAT side enable seamless integration with other systems, allowing for the addition of Critical Work Orders and Supply Chain Work Order data to the database. This feature enables the platform to predict Turnaround Times, considering various factors such as Maintenance Planning Documents, Airframe Work Requirements, customer work orders, and expert insights. The classification of skill sets for findings has improved accuracy. Inconsistencies in maintenance history within the Enterprise Resource Planning system have been identified, and aircraft age remains undefined within the system..

[Audio] The enhancements made to the Estima AI platform include the integration of new data sources such as CWO data into the database, enabling more accurate predictions and analyses. Additionally, the addition of fields to capture actual, predicted, and quoted values in the estimate screen allows for a more comprehensive understanding of the estimation process. Furthermore, the incorporation of CWO data into the estimates enables more precise predictions of turnaround times and material consumption costs. The implementation of access controls ensures that users can only view and interact with relevant data, improving overall efficiency and accuracy. The development of customized MIS reporting templates and the creation of a dashboard, Estimate Navigator, further enhance the user experience and enable more effective comparisons between RFQ inputs and executed data. These enhancements aim to improve the accuracy and reliability of the Estima AI model by comparing its outputs with executed data and implementing improvements based on those comparisons. By doing so, the platform becomes more efficient and effective in supporting the estimation needs of users..

[Audio] The key feature enhancements implemented in this version of the platform include improved probability logic, enhanced spare cost updates, and the ability to identify new findings during estimation. These enhancements are aimed at improving the overall efficiency and accuracy of the estimation process. The system now includes features such as package match percentage displays, multiple part selections, and updated findings lists with cost and man-hours information. Users can also generate Excel summary reports and access historical data through the History tab. These improvements are designed to enhance user experience and streamline the estimation process..