BUILD — Instructional Design Thought Partner
→
Before You Begin
B.U.I.L.D. guides you through purposeful, ethical AI design — and generates a ready-to-use prompt for Gemini, ChatGPT, or Claude at the end.
B
Begin with Purpose
Define the real problem before touching any technology.
U
Uphold Ethics
Design for equity, privacy, and safety from the start.
I
Iterate Collaboratively
Build a practical idea, then push it further.
L
Learn Transparently
Make the AI's role visible to everyone it affects.
D
Develop Responsibly
Turn your thinking into a ready-to-use prompt.
Let's find your starting point.
Two quick questions — no wrong answers. Your responses calibrate the depth of guidance you'll receive at each step.
One more question
⚠ Choose your experience level to begin.
1
Begin with Purpose
Step 1 of 5
What learning challenge do you want to tackle with AI?
Start with the real problem — not the technology. Describe what's not working, who it affects, and what you wish existed.
Your challenge, goal, and who it's for grade, subject, and what's not working
"I teach 10th grade Biology. My students do lab write-ups every two weeks but by the time I give feedback, we've moved on. They never get a chance to revise. I need something that can give them initial feedback on their data analysis section before I read it — not to replace my feedback, but so they arrive at my comments with more self-awareness."
Name the grade, subject, and the specific moment the problem occurs. Include any relevant context about your students — device access, language background, mixed ability levels.
Whose voice shaped this need? select all that apply
It's completely okay to say you identified this yourself — the point is to be aware of whose perspective is and isn't represented yet.
⚠ Describe your challenge and select at least one voice to continue.
2
Uphold Ethics
Step 2 of 5
What are the ethical stakes of your tool?
The time to think about risk is before you design, not after. Three focused questions — privacy, inclusion, and limits — will shape how your resource protects every student it touches.
Privacy and data what student information is involved, and who can see it?
"Students will be pasting their own writing drafts into the tool. This is sensitive — a student's writing can reveal personal experiences, mental health, family situations. I need to ensure nothing is stored after the session ends, no student names appear in the tool, and I've checked with our district tech coordinator."
Think beyond passwords. What does the AI actually see? Could a student accidentally share something personal? Have you checked your district's AI use policy?
Who is this for? select the primary audience — choose all that apply
"My students with IEPs and my English language learners need to be centered in this design, not bolted on. That means the tool must work with screen readers and cannot assume academic vocabulary. I also need to watch that the AI doesn't implicitly reward a particular writing style that reflects only Western academic norms."
Ask: whose needs would expose the worst failures in this tool? Design for them first. Then ask: whose standards does this AI treat as 'correct' — and is that fair to all your users?
Select all that apply. Then use the field below to describe specific individuals or groups whose needs should shape the design — and note any AI assumptions that could disadvantage them.
What should your resource refuse to do? hard limits — these hold no matter what a user asks
"This resource must refuse to: write or complete any part of a student's work for them, discuss topics outside the scope of argumentative essay feedback, respond to questions about the teacher, school, or other students, and produce any response longer than 200 words."
Guardrails are hard limits — things the tool will always refuse, no matter how the user asks. Think about what a student might try to get the tool to do that you would never sanction.
⚠ Address all fields to continue.
3
Iterate Collaboratively
Step 3 of 5
How could AI actually support your goal?
Start with what you could build this week, then stretch your thinking. The most useful idea usually isn't the first one.
Practical approach something you could build or set up this week
"A Gemini Gem that acts as a 'first reader' for student essay drafts. The student pastes their draft and answers three questions: What argument are you trying to make? What evidence feels strongest to you? Where are you most unsure? The Gem then responds with two specific observations about the draft and one question for the student to consider."
Describe what the AI actually does — what it receives, what it asks or says, and what it produces.
Bigger idea or wish list something more ambitious — even if it's not possible yet
"A tool that shows students two versions of their thesis statement — one that preserves their voice and argument, one that translates it into conventional academic English — and asks them which one they want to develop and why. Long-term, I wish it could remember each student's recurring strengths across the year."
Don't censor yourself. A bigger idea doesn't have to be buildable today — naming your vision shapes the design even when it's a stretch.
How will users interact with this? select the closest match
This shapes the entire interaction model in your output. Choose the closest match — then add any context that makes it more specific to your situation.
How will you know if this is working? name the signals that would tell you to keep going — or stop and revise
"I'll know this is working if my most hesitant writers start the revision process without needing me to sit with them. If they can identify one specific thing to change, that's a win. I'll know to stop and revise if students are using the tool to get validation rather than feedback — or if anyone tells me the questions feel confusing or unfair."
Success signals are specific and observable — not 'students like it' but 'students do X differently.' Revision triggers are the moments you'd know something needs to change.
⚠ Describe your practical approach, choose an interaction format, and describe how you'll know if it's working.
4
Learn Transparently
Step 4 of 5
How will users know what the AI did?
Transparency is how you build trust — with students, families, and colleagues. Two questions: how you'll make the AI visible, and where it ends.
How will you make the AI's role visible? disclosure to students, families, and colleagues
"Every screen in the tool will include a visible header: 'This is an AI-assisted tool designed by Ms. Rivera to support your writing process. It does not grade your work and does not report to your teacher.' I will also explain the tool verbally before first use."
Disclosure should happen in multiple places, not just once. Think about: where it appears on screen, what it says in plain language, and whether parents and admin are aware.
What should this tool never replace — and how could it be misused? limits and worst-case scenarios, together
"This resource should never replace a one-on-one writing conference — it can't read tone, emotional state, or the full context of a student's life. Misuse scenarios I need to design against: a student pasting someone else's essay, asking it to rewrite their draft, or using it to generate text to submit as their own."
What a tool won't do and how it might be abused are two sides of the same coin. Naming both here strengthens your guardrails in the output.
Have you checked your school or district AI use policy?
This doesn't block you from designing — but it will appear as a reminder in your output so you don't deploy before you're ready.
⚠ Fill in both fields and select your policy status to continue.
5
Develop Responsibly
Step 5 of 5
Choose your output format.
In design thinking, this is the Prototype stage — you're handing your thinking to an AI to build from. A specific, well-structured prompt means the AI needs fewer exchanges to get it right. Less back-and-forth means less compute, less energy, and a tool that works more reliably for your students from the start.
💎
Gemini Gem Instructions
A Gem is a custom AI assistant you configure inside Google Gemini. You give it a name, a role, and instructions — then share a link with users. They chat with it directly. No coding, no website. Best for conversational tools: writing feedback, Socratic questioning, study support.
⚡
Interactive Resource Prompt
This generates a prompt you paste into Gemini, ChatGPT, or Claude — and the AI builds you a complete, standalone web tool (a single HTML file). Users open it like a website. It has screens, buttons, and a designed flow. Best for structured, step-by-step experiences where you want full control over what users see and do.
📊
Advanced Gem + Sheets
Everything in Gemini Gem Instructions, plus a Google Apps Script that connects your Gem to a Google Sheet. The Gem can pull in class context from your sheet, and student interactions get logged for your review. Best for educators who want a feedback loop — seeing what students are asking and how the tool is responding.
⚡📊
Interactive Resource Prompt + Sheets
Everything in the Interactive Resource Prompt, plus built-in Google Sheets logging. The generated web tool includes a configurable sheet connection so you can review how users interact with it. Best for advanced educators who want both a designed interface and a data feedback loop.
Suggested for you
Loading recommendation...
✓
💎
Gemini Gem Instructions
A ready-to-paste system prompt for a custom Google Gemini Gem — a focused AI assistant with a specific role, ethical guardrails, and an opening message for your users.
⟶ Paste into Google Gemini Gems
✓
⚡
Interactive Resource Prompt
A detailed prompt to build a standalone interactive resource — a web-based tool your users can open and use, with no coding experience required from you.
⟶ Paste into Gemini, ChatGPT, or Claude
✓
📊
Advanced Gem + Sheets
A sophisticated Gem prompt with Google Sheets integration — pull in class context and log student interactions for review. Exact Apps Script code included.
⟶ Paste into Gemini Gems + Google Sheets
✓
⚡📊
Interactive Resource Prompt + Sheets
Everything in the Interactive Resource Prompt, plus a Google Sheets connection — log how users interact with the tool and review responses in a structured sheet.
⟶ Paste into Gemini, ChatGPT, or Claude + Google Sheets
Want to connect this to Google Sheets? — optional, but powerful
▾
What this does
This optional step lets your Gem read from a Google Sheet and write responses to a separate sheet for your review. No external tools. No API keys. Just a Google Apps Script built into Google Sheets.
How to set it up
1
Open your Google Sheet. Go to Extensions → Apps Script.
2
Delete any placeholder code. Paste the script below exactly as written.
3
Click Save, then click Run → run. Approve the permissions when prompted.
4
A new sheet tab called AI_Log will be created automatically.
5
Copy the spreadsheet URL and replace [YOUR SHEET URL] in your Gem instructions.
The Apps Script — paste this into your sheet
function logStudentInteraction(studentName, prompt, response) {
var ss = SpreadsheetApp.getActiveSpreadsheet();
var log = ss.getSheetByName('AI_Log');
if (!log) {
log = ss.insertSheet('AI_Log');
log.appendRow(['Timestamp', 'Student Name', 'Prompt', 'AI Response']);
log.getRange(1,1,1,4).setFontWeight('bold');
}
log.appendRow([new Date().toLocaleString(), studentName, prompt, response]);
}
function getClassContext() {
var ss = SpreadsheetApp.getActiveSpreadsheet();
var sheet = ss.getSheets()[0];
var data = sheet.getDataRange().getValues();
return JSON.stringify(data);
}
function onOpen() {
SpreadsheetApp.getUi()
.createMenu('BUILD Resource')
.addItem('View AI Log', 'viewLog')
.addToUi();
}
function viewLog() {
var ss = SpreadsheetApp.getActiveSpreadsheet();
var log = ss.getSheetByName('AI_Log');
if (log) {
ss.setActiveSheet(log);
} else {
SpreadsheetApp.getUi().alert('No log yet — run your Gem first.');
}
}
⚠ Choose an output type to continue.
Before you share this with your audience
1
Read the opening message out loud.
AI-generated language can feel formal or cold. If it doesn't sound like something you'd actually say, revise it before you share it.
2
Try to break it before your audience does.
Test with 5–6 prompts, including some that try to take it off-topic. If it wanders, go back and tighten your guardrails.
3
Use it as your audience would — start to finish.
Don't share anything you haven't experienced yourself. Test with the kinds of responses your most hesitant users would actually give.
4
Tell your audience exactly what this is and what it isn't.
An introduction is not optional. Users need to know it's AI, who designed it, and what it will never do.