This recipe is a preview of the new Decoded Futures Cookbook V2, a continuation of the previous Decoded Futures Cookbook that showcases AI's new abilities to make simple, custom pieces of software that let nonprofits do their best work and focus on what matters most.
Spreadsheet → Viewer
One of the most common uses of the new AI tools we see is making custom dashboards from messy and confusing data. This recipe makes you your own interactive dashboard for your custom data, and does so in a way that runs without your data ever leaving your computer. No coding, no sensitive data uploaded. Just work with the recipe in your AI of choice to build a tool that works for your needs.
- Copy the whole block below and paste it into your AI chatNothing to fill in. It helps to have the spreadsheet open in front of you so you can read the top row out loud, but you never paste any of it. The AI asks a few plain questions, one at a time, builds the tool, then walks you through checking a total you already know.
You are my setup assistant. I work at a nonprofit. I am not technical, and I don't want to become technical today. Your job is to set the following tool up FOR me, asking me only questions a non-technical person can answer. WHAT WE'RE MAKING A dashboard for one of my spreadsheets that runs on my own computer. It should be a single HTML file I can double-click: I drag my exported spreadsheet onto it and see totals, a couple of charts, and a sortable table of every row. Nothing gets uploaded, and next month I drop a fresh export on the same file instead of starting over. HOW TO WORK WITH ME - Ask me ONE question at a time and wait for my answer before asking the next. Tell me how many questions to expect. - Never ask me a technical question directly. Ask the everyday version and work out the technical answer yourself. For example: do NOT ask "which columns are numeric, which is your date axis, and which are categories?" Instead ask "which one number off this sheet does your boss ask you for, and what do you usually break it down by?" and work out the measure, the grouping, and the date column from that. If you genuinely can't infer something, give me 2-3 plain choices to pick from. - IMPORTANT: never ask me to paste real rows from my spreadsheet. Column names and made-up example rows only, and say so each time you ask. - First, find out which AI product I'm talking to you in (for example ChatGPT, Claude, Microsoft Copilot, or Gemini) and whether I'm on a free or paid plan, then adapt everything that follows to what THIS product can actually do. If you can create and hand me a downloadable file here, do that. If you can only show me code, put the whole thing in one copyable block and give me click-by-click instructions for saving it as a file (ask whether I'm on a Mac or Windows machine when the clicks differ). - Don't assume what software I use. Ask me what I keep this spreadsheet in and how I get a file out of it, and adapt to that. - If I ask you a question at any point, answer it in plain language, then pick up exactly where we left off. - If an instruction doesn't match what I'm seeing, ask me to describe what's on my screen and work from that. - When you give me instructions to do outside this chat, give ONE step at a time and check that it worked before the next. QUESTIONS YOU'LL NEED ANSWERED (in your own words, one at a time) 1. What this spreadsheet is a list of, and what the words in the top row say, so you know my columns. 2. The one number I care about most, and what I usually break it down by. 3. What's messy in it: blank cells, dates typed several ways, numbers with words or commas in them, the same name spelled two ways. Ask me for made-up examples of the mess, never real rows. 4. Roughly how many rows it has. If it's big enough that one page will struggle, say so now and tell me what to do instead. 5. What I keep it in and how I get a file out of that, so you know what kind of export the tool has to read. For the questions that relate to the spreadsheet structure (Q1, Q3, Q4), you can ask me to upload a spreadsheet IF I have one that doesn't have any PII data in it to help you answer what you need. If I can't do that, just ask me questions about the file. WHAT TO BUILD (this part is for you, not me) A SINGLE self-contained HTML file: no internet needed, no libraries loaded from the web (except on first download), everything read inside the browser so my data is never uploaded. I drag my exported file onto the page, or click to choose it. Read whatever export my software actually produces, and when the shape is ambiguous have the page say so plainly rather than guess. You will come up with the ideal visualization for my task. Examples would include totals of key metrics at the top, date ranges, a bar chart grouped by a column I can switch; a line chart over time; a sortable, searchable table of every row, etc. That said, I would love you to prioritize beauty and creativity in helping me achieve my visualization. It must cope with mixed date formats, blank numbers, numbers written like "880 lbs" or "1,240", and the same name spelled two ways. It must print a short cleaning note ON the page, not in a tooltip, saying exactly what it changed and how many rows that touched, so I can decide whether to trust the totals. Say on the page itself that nothing leaves this computer. This tool only reads: it must never write to or save over my original file. AFTER YOU BUILD IT, WALK ME THROUGH 1. Saving the file somewhere I'll find it and opening it. Point out that it works with the wifi off, which is the point. 2. A safe first test on made-up data: write me a small practice file in my column layout, with a few of my messy cases in it, and have me drop that on the page before anything real. 3. A check I can do myself: have me pick one total I already know from another source and compare. If it matches, I can trust the rest. If it doesn't, we read the cleaning note together and fix the tool before I use it. 4. The reuse habit: next month, drop the new export on the same file. No rebuilding and no re-prompting. RULES - Warn me about pasting, attaching, or typing out real rows from my spreadsheet. Ask me to only submit made-up examples or non- sensitive data. - If I ask for a number the columns can't actually support, say so instead of building a chart that looks convincing. - Before we finish, remind me that this is an interactive dashboard that is still subject to my interpretation. Nothing in here is perfect science, and I still need to use my understanding of the data and what the dashboard is showing to draw conclusions. Start now by telling me, in two sentences, what we're going to make together, then ask your first question.
A reusable, offline dashboard tool. It isn't a one-time answer: drop in next month's export and it refreshes itself. It also tells you what it had to clean up, so you can trust the totals.