Class Introduction
This was an AI training session led by Dinesh focused on enterprise applications of artificial intelligence, specifically covering Excel sheet forecasting and machine learning models. Dinesh explained how to use Copilot in Excel to create tables, generate graphs, and perform predictive analysis using various machine learning models, emphasizing the importance of understanding R-square values and feature engineering. The session covered the three domain areas of AI implementation: personal productivity, enterprise AI applications, and AI agent creation. Dinesh introduced the concept of AI agents and their technical architecture, discussing model selection (GPT-5 Chat, GPT-5 Reasoning, Claude, etc.) based on business needs and cost considerations. He explained the importance of proper instructions and knowledge sources for AI agents, and mentioned that the next session would focus on practical AI agent creation using Copilot Studio. The session concluded with Dinesh promising to share documentation and resources with participants, including a PDF guide on AI agent configuration and model selection for the Microsoft AI Transformation Leader certification.
AI Program Roadmap and Applications
The meeting focused on providing clarity about the program roadmap and enterprise applications of AI tools. The facilitator explained the importance of prompt engineering, using notebooks, and customizing AI tools for specific objectives, referencing a website called "there is an AI for that" that lists AI tools by business verticals. The discussion covered Copilot 365, which learns from company knowledge sources like Outlook and SharePoint, and the importance of creating specific notebooks for different roles and objectives, with proper instructions for the LLM to function effectively.
Microsoft Copilot Notebook Implementation
The team discussed technical issues with screen sharing before focusing on personal productivity notebooks in Microsoft 365. Dinesh explained how Copilot notebooks work by combining co-pilot instructions with knowledge sources, using personal finance and DAX formula examples to demonstrate the concept. The discussion revealed that participants needed more practice with applying Copilot in Excel, as they hadn't tested the functionality with provided datasets despite previous classes on the topic.
Excel Copilot Implementation Discussion
The team discussed using Excel Copilot, focusing on its ability to create tables and generate visualizations. Ahmed explained that Python can be used as a backend for various data analyses, including predictive and descriptive analysis. Amr requested the prompts be shared on WhatsApp for easier practice, and the team agreed to do so. The discussion touched on creating tables, templates, and graphs, with an example of a staff roaster schedule.
Related Offerings
Risk Register Data Visualization
The team discussed creating graphs and charts for risk registers, focusing on both qualitative and quantitative data analysis. The instructor explained how to use Python Word Cloud functionality within Excel to analyze qualitative data, particularly demonstrating how to visualize word frequency in risk assessment data. The discussion emphasized that while Excel natively supports quantitative charts, qualitative analysis requires backend Python processing, and highlighted the importance of aligning risk dashboards with ISO 31000 ERM standards.
Risk Register Dashboard Planning
The team discussed creating a risk register dashboard using Excel with Copilot, focusing on three key objectives: creating tables/templates, converting data into information through charts and graphs, and performing predictions and forecasts. They explored predictive analysis techniques, including the use of machine learning models and R-square values for forecasting financial data like balance sheets. The discussion explained the foundations of machine learning models and their applications in Excel, emphasizing that different models are suitable for different data patterns and trends.
Machine Learning for Financial Forecasting
The team discussed implementing machine learning models for financial forecasting, with a focus on using different models for each line item based on their specific data patterns and historical trends. The speaker explained that AI models provide probability-based predictions rather than 100% accurate forecasts, with R-square values indicating confidence levels. They demonstrated how to apply various machine learning models including linear regression and gradient boost to different financial components, showing that each model produces different confidence levels (ranging from 81% to negative values). The discussion concluded with confirmation that these predictive models can be directly applied to budget forecasting, with specific examples given of how similar approaches are used in industries like airlines for price prediction.
Enterprise AI Applications Discussion
The team discussed the application of machine learning models in business scenarios, emphasizing that these models can be applied across various domains beyond just financial analysis. They explored enterprise AI applications, including using AI agents to automate tasks in risk management, ERP systems, and other business processes. The discussion covered the creation of AI agents using tools like Copilot Studio, which can perform data processing, predictions, and real-time interactions with systems through Model Context Protocol (MCP). The instructor announced a shift in focus to enterprise applications, moving away from individual AI tools and prompt engineering, and invited questions about the program roadmap before proceeding with the next topic on AI agents.
AI Agent Configuration Training
Dennis explained how to create and configure AI agents, including setting names, icons, and technical configurations. He emphasized the importance of proper initial setup and mentioned that creating professional AI instructions requires significant effort. Dennis assigned an Excel-based forecasting and machine learning task related to risk register analysis as part of the Microsoft Certified AI Transformation Leader certification, which costs $50.
AI Model Selection Discussion
The team discussed the selection of AI models for different business needs, with Dinesh explaining that the choice depends on efficiency, effectiveness, and specific requirements like reasoning versus data retrieval. Dinesh emphasized that GPT-5 Chat is suitable for simple information retrieval while GPT-5 Auto can perform both reasoning and chat functions, and Claude models are better for qualitative analysis but more expensive. The discussion concluded with Dinesh recommending the use of multiple models based on specific goals rather than using a single model for all purposes, and he planned to provide a detailed review of Claude in a future class.
AI Certification Requirements Discussion
The team discussed the requirements and costs for achieving Microsoft certification using AI agents. It was clarified that the Team Academy account can be used for free to create AI agents. The instructor explained the importance of selecting appropriate models and giving specific instructions to AI agents, emphasizing cost-benefit analysis and business case studies. Amr requested documentation on AI models and their applications, which the instructor agreed to share, including a consolidated PDF guide on AI Transformation Leader Certification. The next class will focus on practical AI agent creation starting at 2:30.
Ready to move beyond AI tools and discover how artificial intelligence can transform everyday business processes?
👉 Explore our AI Transformation Training Program and learn how to apply Copilot, machine learning, predictive analytics, AI agents, and enterprise AI solutions to real-world business challenges.
