I need a small internal GPT-powered assistant that can read a library of SIAM Excel workbooks – one file per month from 2020 through 2025, each listing every automotive brand and its units sold – and instantly answer written questions such as: • “Show the sales trend for Brand X from 2021 to 2023.” • “Which brand outsold Brand Y in March 2024?” • “Give me the monthly totals for Brand Z in 2022.” Users will interact through a simple text box; no voice or drop-down menus are required. As the admin I only want to drag-and-drop the new spreadsheets into a secure backend folder; the agent should pick them up automatically without any code changes. We run Microsoft 365, but SharePoint access for GPT is restricted in GPT agent. I have uploaded the SIAM files for sample as well. Key expectations – Automatic ingestion and indexing of every new monthly file. This will be like, as soon as latest month data is released, i will upload it somewhere. – Natural-language querying that returns precise numeric answers or small pivot-style tables. – Support for the three core question types: sales trends over time, brand-to-brand comparisons, and single-brand monthly figures. – Clear error handling when a query references a date or brand that is not in the data. Deliverables A Prompt and process of how to make it possible. but need to create a GPTAgent. Not a Agentic AI thing Acceptance criteria • Queries typed in plain English return correct results within five seconds. • Adding a brand-new monthly workbook triggers no crashes and the fresh data is immediately searchable. • All calculations match manual checks in Excel. If any commercial libraries or OpenAI features carry extra costs, please highlight them up front so I can clear them with finance.