# Using the FranchiseDepth database with AI assistants This folder works with any assistant that accepts file uploads. Pick the files that match what you want to do, upload them, and give the assistant the instructions in `AI_INSTRUCTIONS.txt` (paste it as your first message, or into the instructions box of a custom assistant). ## Which files to upload | You want to... | Upload | |---|---| | Ask quick questions about fees, earnings, closures or growth across brands | `brand_summary.csv` | | Read about specific brands in one industry | the industry's file in `profiles_by_industry/` | | Run deeper analysis (the assistant writes code) | `franchisedepth.sqlite`, or the files you need from `csv/` | | Look up every column and method | `AGENTS.md` | Upload limits depend on each tool and plan. If a file is too big, start with `brand_summary.csv`, which is small, or one industry's profile file. ## ChatGPT - **One chat:** attach `brand_summary.csv` (and any `csv/` files you need) with the paperclip, then paste `AI_INSTRUCTIONS.txt` as your first message. - **Your own GPT** (on plans that offer GPTs): create a GPT, paste `AI_INSTRUCTIONS.txt` into Instructions, upload `brand_summary.csv`, `AGENTS.md` and the industry profile files you want as Knowledge, and turn on the code/data-analysis capability. - ChatGPT's data analysis can often open `franchisedepth.sqlite` with Python. If it can't, use the CSV files instead. ## Claude - **One chat:** attach `brand_summary.csv` or a profile file and paste `AI_INSTRUCTIONS.txt`. - **A Project:** create a project, paste `AI_INSTRUCTIONS.txt` into the project instructions, and add `brand_summary.csv`, `AGENTS.md` and profile files to the project knowledge. - **Claude Desktop with live SQL:** connect the whole database with the MCP connector (steps in `README.txt`). Claude can then query every table directly. ## Gemini - **One chat:** upload `brand_summary.csv` (CSV and Excel files are accepted; several files can go in one prompt) and paste `AI_INSTRUCTIONS.txt`. - **A Gem:** create a Gem, paste `AI_INSTRUCTIONS.txt` into its instructions, and add `brand_summary.csv` and profile files as its knowledge. - **Google Sheets:** import `brand_summary.csv` into Sheets and use Gemini in Sheets on it. ## Grok - Attach `brand_summary.csv` (and a profile file if you like) with the paperclip, and paste `AI_INSTRUCTIONS.txt` as your first message. ## Microsoft Copilot - Upload `brand_summary.csv` in a Copilot chat and paste `AI_INSTRUCTIONS.txt`. - With Copilot in Excel, open `franchisedepth.xlsx` and ask about the tabs directly. ## Perplexity - Create a Space, add `brand_summary.csv` and profile files, and paste `AI_INSTRUCTIONS.txt` into the Space's instructions. ## Coding assistants (Cursor, Claude Code, Codex, Gemini CLI, GitHub Copilot) - Open this folder as the project. They read `AGENTS.md` (and `CLAUDE.md` / `GEMINI.md`, which point to it) automatically. Ask them to query `franchisedepth.sqlite`. ## Tips for good answers - Ask for a table: "Compare these five brands' investment, royalty, closure rate and comparable revenue in a table." - Ask for sources: "Give the FDD year for each figure and the page for Item 19 figures." - Ask it to say when something isn't disclosed, rather than guess.