Six tracks
Pick one person. Solve one day in their life.
Each brief is a whole industry. Your job is to shrink it to one real person and one real moment, then build something they could actually use.
TRACK 1 · HEALTH & WELLNESSRoom AA 201B
The appointment nobody keeps
People miss doctor and clinic visits they already booked. Usually they did not forget. The bus takes too long, the work shift changed, the forms are in the wrong language, or nobody can watch the kids.
How might we help one person actually make it to the care they already scheduled?
Start here
Who do you know who has missed an appointment? What really stopped them? What would they have needed to hear, and when?
Where AI can work
- Understanding a messy text reply in any language
- Matching a barrier to a fix: a ride, a new time, a video visit
- Pointing someone to the right service on the first try
You could build
- A reminder assistant that asks “anything in the way?” and offers options
- A navigator: “here is what I need” becomes where to go and what to bring
- A mockup of a clinic screen showing who needs a call today
Think about (this is scored)
Health information is private; who sees it? Your tool is not a doctor and must never give medical advice. What about someone with no smartphone?
TRACK 2 · LOGISTICS & SUPPLY CHAINRoom AA 201C
The last mile is the hard mile
Things cross an ocean on time and then get stuck in the last few miles to a store or a doorstep. Giant carriers have software for this. The florist, the food bank, and the two-van delivery company have a whiteboard and a group chat.
How might we help one small operator plan like a big one?
Start here
When did a late or missing delivery cost you or your family something? Who was making the plan that day? What did they not know in time?
Where AI can work
- Putting stops in a sensible order and explaining why
- Guessing next week’s demand from last month’s numbers
- Turning tracking jargon into one clear sentence and a next step
You could build
- A delivery planner: paste today’s stops, get a plan and the customer texts
- An order suggester for a small shop or food pantry
- A “where is it stuck?” tracker mockup
Think about (this is scored)
Faster routes can mean more pressure on drivers; where are the breaks? If the AI is wrong, a truck goes to the wrong place, so how does someone check it? Location data is sensitive.
TRACK 3 · SMALL BUSINESSRoom AA 204
Competing without a marketing department
The owner of a small shop is expected to post on social media, answer reviews, run ads, and still do the actual job. Big chains have whole teams for that. The owner has ten minutes after closing.
How might we give one local business owner an hour of their day back?
Start here
Pick a real business you know: a taqueria, a barber, a nail salon, a market. What is the owner doing at 9 PM that is not their real job?
Where AI can work
- Drafting posts, captions, and flyers in the owner’s own voice
- Replying to customers and reviews in English and Spanish
- Spotting what sells and suggesting a special
You could build
- A one-week content kit for your chosen business
- A reply assistant the owner approves with one tap
- A simple advisor that reads a sales sheet and says what to promote
Think about (this is scored)
No fake reviews and no made-up claims. The owner approves everything before it goes out. Does it still sound like them?
TRACK 4 · EDUCATIONRoom AA 205
The students who quietly disappear
A student stops showing up, and weeks pass before anyone who could help notices. They usually had a problem that could be fixed. They did not know who to ask.
How might we reach one student before they disappear, in a way they would actually answer?
Start here
Think of someone who faded out of a class or school. What was the first sign? How would they have wanted someone to reach out?
Where AI can work
- Reading a short reply and picking the right help: tutoring, aid, food, counseling
- Writing a message that sounds human
- Turning “where do I go for this?” into a real office and real hours
You could build
- A check-in assistant that hands the student to the right person
- A campus navigator for new students
- A weekly summary for a counselor, with a drafted message for each student who went quiet
Think about (this is scored)
Being labeled “at risk” can follow a student. Who sees that label? Can the student? Does your outreach feel like care or like being watched?
TRACK 5 · CLIMATE & ENERGYRoom LL 102
Heat lands unevenly
On the same afternoon, one neighborhood can be far hotter than another a few miles away. Shade, pavement, and housing decide it. The people with the least cooling often hear about help last.
How might we get relief to one person on the hottest day of the year?
Start here
Where is the hottest place you regularly spend time, and why is it hot there? On the worst day last summer, who near you was most at risk?
Where AI can work
- Matching a resident’s situation to nearby relief, in their language
- Describing and scoring photos of a block for shade
- Organizing evidence into a case someone with a budget will read
You could build
- A heat-day helper that gives a plan for the day
- A shade audit of blocks you know, ranked by need
- A one-page ask to a city office for one specific fix
Think about (this is scored)
The people most at risk may have no smartphone, no car, and limited English. A map of “risky blocks” can stigmatize a neighborhood. Never point to a vulnerable person’s home.
TRACK 6 · CREATIVE MEDIARoom MM 122
Getting heard in an infinite feed
Anyone can post. Almost nobody gets seen, especially people telling stories from their own community with no budget and no team.
How might we help one creator or one community story reach the people who need to see it?
Start here
What story from your community deserves a bigger audience? Who needs to see it, and where do they scroll? Which part of making it takes the longest?
Where AI can work
- Speeding up the slow parts: scripts, shot lists, captions, alt text
- Turning one story into several formats
- Researching who the audience is and what they already watch
You could build
- One real story in three formats with a posting plan
- A creator’s assistant: rough idea in, shot list and hook options out
- An audience brief recommending an angle nobody else is taking
Think about (this is scored)
Did the people in the story say yes? No cloning real people’s voices or faces. Say when AI helped. Keep the voice that made the story worth telling.