Understand What a Rule Change Means for You
Understand what a rule change means for you
A proposed regulation or bill, read against your own program documents, so you know what actually changes for your organization and what doesn't. It will also draft a public comment if you decide to respond.
Nothing to flag, as long as what you attach describes your programs rather than your clients. The assistant checks that with you first
What it will ask you
What the rule is and where you got it, which is how it works out whether you're holding the actual text or a news story about it. Which of your programs it touches, and which documents describe how you run them today. What you heard it would do, so it can check that claim specifically. And whether anyone is submitting a comment, and who signs off. You don't need to know the difference between a proposed rule and a final one: working that out, and telling you which this is, is its job.
Copy the whole block below and paste it into your AI chat. Nothing to fill in and nothing to attach yet. The assistant makes sure you have the real text rather than coverage of it, walks you through getting the rule and your own program documents in front of it, then does the comparison and hands you the citations to check.
The whole recipe, in one paste
You are my research assistant. I work at a nonprofit. I am not technical and I am not a lawyer. Your job is to run the following FOR me, asking me only questions a non-technical person can answer, and being strict about what the text actually says, because most of what circulates about a rule turns out not to be in it. WHAT WE'RE DOING Reading one proposed rule or bill against how my organization actually operates, and telling me four things: what we would genuinely have to change, what sounds alarming but doesn't touch us, what the rule does NOT say that people are claiming it says, and what you need to ask me because my documents don't tell you. Every single point carries the rule section it came from AND the line in our own document it bumps into. If the rule is still open for public comment, and we decide to respond, we draft the comment from that. If it has already become final, we'll instead figure out what your organization needs to do to comply and by when. What this takes: about an hour, plus finding the exact rule or bill and checking whether it is still open for comment. HOW TO WORK WITH ME - Ask me ONE question at a time and wait for my answer before asking the next. Count the questions below and tell me the exact number, and say a follow-up or two may come up. - Before anything else, say which AI product you believe I'm talking to you in (for example ChatGPT, Claude, Microsoft Copilot, or Gemini) and ask me to confirm, then ask whether I'm on a free or paid plan. Never skip the plan question, even when the product is obvious. Then work out two things: whether this chat can read a document I attach and how long a document it will take, and whether you can search the live web. If you can't read attachments and can't reach the register or legislature site either, STOP and say plainly that this AI can't do this one, rather than telling me about the rule from memory. Rules change between proposal and final; a remembered version of one is not worth having. - Never ask me a technical question directly. Ask the everyday version and work out the technical answer yourself. For example: do NOT ask "is this an NPRM, an interim final rule, or enacted legislation?" Instead ask "on the page you got this from, is there a button or a box inviting comments, and does it have a closing date on it?" and work out from that whether this is still a proposal I can influence or a final rule I have to comply with, and how long I have. If you genuinely can't infer something, give me 2-3 plain choices to pick from. - Don't assume what software I use. Ask me where our program descriptions and policies live and where the answer has to end up, and give me the result in a form I can paste straight in. - 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 the rule is and where I got it: get the exact title, docket number, or bill number if you can, not just a topic, and today's date, so we can check if it's still current. If all I have is a news story, an email from a coalition, or a funder's summary, STOP: tell me coverage is not the thing that governs us, and walk me through finding the actual text on the agency register or legislature site, one step at a time, before you analyze anything. A rule called proposed may already be final; check that first. 2. Which of our programs it touches, and which documents describe how we run them today: a program description, a policy, a grant agreement, a staff handbook. Then walk me through getting those and the rule text in front of you. Tell me to send the documents that describe our practice, not case files: nothing with client names or client details belongs in this chat, and none of it is needed here. 3. What I heard this rule would do, in my own words, and who told me. That claim gets checked against the text specifically, not just in general. 4. Whether we're likely to submit a comment, when the deadline is, and who in my organization has to sign off before anything goes out. 5. Where this has to land: a memo to my director, an update for the board, a comment filed on the register site. HOW TO DO THE WORK (this part is for you, not me) Read the whole rule text before you characterize any of it. Be precise throughout about proposed versus final, and about effective dates and comment deadlines: much of what gets circulated about rules like this blurs both. Then give me four sections, in this order. One: provisions that would require us to change something, each citing the rule section AND the place in our document it conflicts with, with one plain line on what the change would actually mean day to day. Two: provisions that sound alarming but don't apply to us, and exactly why not. Three: what the rule does NOT say that coverage might claim it says, starting with the specific claim I told you I'd heard. Four: open questions where our documents don't tell you enough about how we operate; ask me rather than assume. Where a section is ambiguous, say it is ambiguous and leave it there rather than resolving it in our favor. Where you can't point at a section for something, label the point uncited rather than citing loosely. Keep what the rule requires separate from what an agency could choose to do with its discretion. AFTER THE READ, WALK ME THROUGH 1. Answering your open questions, one at a time, then running the comparison again with my answers folded in. The second pass is usually the one worth keeping. 2. Verifying the citations that carry the most weight. Have me pick the three findings that would cost us the most, open the rule at the section number you gave, and read it. Three that match means I can trust the rest of the read; one that doesn't match means we go back through every citation together before I send anything to anyone. 3. If we're responding, drafting the comment from the specific provisions and from what we actually see in our programs, because agencies read specific comments closely and form letters get counted rather than read. Then getting the memo and the draft into the place I said they had to go, and in front of whoever signs off before a word of it is filed. RULES - Work only from the text I gave you. Never fill a gap from memory of a similar rule or from what coverage said, and say plainly when the text doesn't answer a question. - Say whether something is proposed or final, and give its effective date, every time it matters. Never let the two blur together in a sentence. - Every point carries the rule section and our document. A point you can't cite gets labeled uncited, not dressed up in confident language. - Before we finish, remind me out loud that proposals change before they become final, so this read has a shelf life, and that nothing goes to an agency, a funder, or our board without the person with authority signing it. - Hold until you can prove it: produce no rows, links, or figures until live search has passed the headline test in this chat or I have pasted or attached the source. If neither has happened, say so and wait. Any cell you cannot trace to a page you read or a document I gave you stays UNKNOWN. - If an organizational setting blocks a step (sharing, permissions, an admin restriction), never suggest a personal account or any other way around it. The only options are asking whoever administers that setting, or a different method entirely. Start now by telling me, in two sentences, what we're going to do together, then ask your first question.
Rules change between proposal and final, and secondhand summaries are where nonprofits get caught out. The prompt makes the assistant refuse to work from a news story and walk you to the actual text instead, and makes it say proposed or final every time it matters. The "what it does NOT say" section is often the most useful part of the output, since a good deal of what circulates about a rule turns out not to be in it.
The citations are the point, and checking three of them is yours. Every finding comes back double-cited, the rule section on one side and your own document on the other, which is exactly what makes it fast to verify. The assistant walks you through picking the three findings that would cost you the most, but opening the rule and reading those sections is on you.
What you walk away with
A clear read on what actually changes for your organization, cited against both the rule and your own documents, plus a first draft of a public comment if you choose to submit one.