Day 5·Design an AI solution
Build Your AI Solution
Granola Notes — posted after class9:00–10:30 · Morning Session: Know Your Project
Why This Morning Matters
Before you can build something, you need to understand what you're building and why. This morning we're going to make sure everyone actually understands their own project — not just the words in the doc, but what it means and why it matters.
Step 1 · Laptops Closed (5 min)
No screens. Each team gets one or two minutes to answer out loud:
- What problem are you solving?
- Who has this problem?
- What does your solution actually do?
Don't worry if this is hard. That's the point.
Step 2 · Read Your Own Work (25 min)
Open your team's combined project doc from the IDEA Student Content folder.
Read everything your team has produced across Days 1–4. All of it. Don't skim.
While you read, leave exactly three comments on the doc:
- Something you can explain in your own words — a part of your project you genuinely understand and could describe to a stranger
- Something you're not sure is actually true — a claim, a number, or an assumption that you're not confident in
- Something you think is weak or missing — a gap in your problem statement, your market sizing, your BMC, or your solution idea
These comments are for you, not for a grade. Be honest.
Step 3 · Get Honest Feedback from AI (15 min)
You've read your own work and formed your own opinion. Now ask Gemini to review it critically.
Open a new Gemini conversation, paste your full combined doc, then run this prompt:
Here is all the work my team has produced so far across Days 1–4: [paste the full combined doc] Review everything critically. I want honest feedback, not encouragement. Specifically: 1. Does the market sizing in Part A actually match our problem statement and domain? Or did it drift to a different product or market entirely? 2. Are the numbers in our market sizing real and sourced, or did they get made up? Be honest. 3. Is there anything in the Business Model Canvas that contradicts or doesn't match what we found in the market sizing or problem statement? 4. What is the weakest part of our entire submission — the thing a city official or investor would push back on hardest? 5. What is one specific thing we should fix before we build a prototype? Be direct. Don't soften the feedback.Read the feedback as a team. Compare it to the three comments you left yourself. Did Gemini find something you missed? Did you find something Gemini missed?
Step 4 · Explain It Again (10 min)
Laptops open this time. One question per team:
What is the one thing your solution actually does for the user?
Not the whole project. Just that one thing.
Step 5 · Why This Matters — Vibecoding Demo (20 min)
You can't build something you don't understand. And when you're working with AI to build a prototype, the AI will sound confident whether your idea is good or not. Your job is to know your project well enough to push back on the output, not just accept it.
That's exactly what you just practiced with the feedback prompt — using AI to critique AI output. That same skill is what makes vibecoding actually work.
Here's what the iteration loop looks like in practice.
[Teo live demo — vibecoding a real project, showing: try something → see what's wrong → describe the problem specifically → try again]
The skill isn't writing the perfect prompt the first time. It's knowing what's wrong with the output and being able to say why.
10:30–10:45 · Field Trip Preparation
Before we leave, spend a few minutes preparing. The founders at Newlab are building real physical products — hardware, manufacturing, deep tech. That's different from what you've been building this week, but the same questions apply: who has the problem, who pays, what does it cost to deliver.
Your job during the visit is to listen for the things that match or contradict what you've been learning. Bring at least one question from the list below to ask.
Questions for the Founders
Warm-up / origin
- What problem were you personally frustrated by before you started this company?
- Why did you build a physical product instead of software or an app?
Market and business model
- Who is your first paying customer, and how did you find them?
- What's the hardest part about scaling something that requires factories or physical space, compared to scaling an app?
- Who pays for this — the end user, a government, or a business customer? Why?
Decisions and tradeoffs
- What's one version of this idea you tried and abandoned?
- What's a decision you made early on that you'd make differently now?
[PLACEHOLDER — add any additional questions from class brainstorm before Day 5]
11:05 · Walk to Jasper Kane → Lunch to Go → Bus to Brooklyn Navy Yard
Lunch options at Jasper Kane (6 MetroTech): chicken sandwiches, tuna wraps, falafel wraps, sushi platters, sushi rice bowls, pizza. NYU meal swipe may include a drink and snack.
We walk to Tillary St/Jay St and take the B57/B67 bus to Brooklyn Navy Yard. Arrive by 12:00pm.
12:00–1:00 · Newlab Field Visit
Brooklyn Navy Yard. Founders' presentations, Q&A, and a tour of the Newlab space.
1:45–3:00 · Debrief + Continue Building
[PLACEHOLDER — debrief structure: what did you see, what connects to your project, what changes about your prototype idea after seeing real founders build]
After the debrief, continue building your AI prototype from the morning session.
Field Visit Reflection
Before the end of day, write down your answers to these three questions. There's no right answer — just be honest about what stuck with you.
- Which founder's problem felt most like something you'd personally want to solve?
- What's one thing you saw today that made you want to learn more — something you might look up once you're back home?
- What's the equivalent "hard part" for your own pitch idea? What's one thing you saw today that changes how you think about your own project?
Submit your reflection to [PLACEHOLDER — submission link or instructions].
3:00–4:00 · Resume & Project Experience
NYU Wasserman Career Center coaches will join for training on resume writing and how to present project experience on a resume.
[PLACEHOLDER — any prep or materials students should have ready for this session]