ISB AI Founders
Day 1 · 15:15 · you have the tool

Your first hour as a company.

Three pushes: prioritize, build, polish. I call the transitions; your team does the work. Standup at 16:15.
ISB AI FoundersFOUNDATIONS
Before push 1 · two minutes of theory

Three tests decide what you build first.

1 · The pain test
Does this feature directly kill the pain in your thesis? The ONE that does is your core. It gets built first, before anything else.
2 · The journey test
Walk your user: they arrive, they do something, they get the result. Features on that straight line stay. Detours wait.
3 · The works-without-it test
Would the product still make sense this week without it? Yes means nice-to-have. Cross it out for now, it is not dead, just later.
The order rule: pain-killer first, then what the journey demands, pretty last. That numbered list is your build order for push 4.
Push 1 of 3 · I call the next one
LAPTOP OPEN · one per team

Ask AI for the features. Pick the core ones.

1Run the feature prompt below on your picked solution.
2Run the three tests: circle the ONE that kills the main pain, keep what the journey needs, cross out the rest.
3Number the survivors 1-2-3: your build order. Write the SEE / DO / GET screen sentence for feature 1. Titles: CEO, CTO, CPO.
"Here is my product: [one sentence]. Brainstorm 15 features it could have. Mark which 3 are CORE (the product is pointless without them) and which are nice-to-have. For each core feature: what the user sees and what they can do."
15:00
Build order + screen sentence for feature 1
The translation move · solution → screen → prompt

A solution is an idea. A screen sentence is buildable.

The formula, one sentence per feature
When the user arrives they SEE [   ], they can DO [   ], and they GET [   ].
Solution, as picked
"Let students pre-order lunch so they skip the line."
Screen sentence, buildable
"They SEE today's menu as cards, they can DO one tap to order, they GET a pickup code."
That sentence, word for word, is what you type into the coding agent. Write one per core feature before you build it.
Push 2 of 3 · I call the next one
LAPTOP OPEN · one per team

Build the features. In order.

1Feature 1: type its SEE / DO / GET screen sentence into the agent, word for word. Let it run.
2Test it like a user. Wrong? "I expected [X], it does [Y]." Right? Ask for feature 2.
3Write the screen sentence for feature 2, then build it. Down the list, in order; design comes next push.
30:00
Features live, one by one
Push 3 of 3 · last one
LAPTOP OPEN · one per team

Design pass. Then make sure it's live.

1One design prompt, pick a card below. One, not five.
2Ship check: the URL loads on a phone, or the preview is ready for the big screen.
3Decide who talks at standup: thesis, build order, live demo. 60 seconds.
Style prompt
"Apply a liquid glass design: frosted panels, soft shadows, rounded corners."
Copy a site you love
"Make it look like airbnb.com: same feel, my content."
Or just describe it
"Dark, minimal, big type, like a modern startup site."
15:00
Live, styled, standup-ready
If you finish early
OFFLINE · screens closed

Interview each other. Make the thesis harder to kill.

1Pair up inside your team. One asks, one answers, then switch.
2Ask the same 4 questions you'll use on real strangers: walk me through the last time this happened. What did you do about it. Have you tried anything else. If it vanished forever, would you actually miss it?
3Any answer that surprises you goes straight onto the spec sheet before you move on to build.
16:10 · five to go

Build orders on the table, screens live. Standup in five.

Whatever you built is what shows. Get the URL loaded, get ready for 60 seconds on the big screen.