Edition 01 · Practice statements
These are practice statements: thirteen open questions to think about, argue over and explore before kick-off.
None of them is a task. Each one points at a frontier, and a dozen teams could take the same question in a dozen directions: an agent, a tool, a protocol, a game, an experiment, an interface, a piece of infrastructure.
They bind no one and earn no extra marks. They are here to start the thinking early.
Explore and plan now. At kick-off on 21 Nov 2026 a separate set of official problem statements is released and the build starts; only work built during the build phase can be submitted.
The best answers are probably ones nobody has thought of yet.
Start anywhere. See where it leads.
Team Homingo Hack Days
Each question sits under its main track. Your team can take any question that fits its track.
More and more of the asking, offering, agreeing and arguing in the world may soon happen between agents before any person sees it.
Language, trust, reputation, money, rules, manners: most of it is still unbuilt for them.
Invent any piece of that world, or find out what goes wrong without it.
Discovery is a loop: notice, guess, test, be wrong, try again. Software can now attempt every step of it.
Close any part of that loop, in any field, and see what it turns up.
A new way of searching counts as much as a new result.
Most agents exist for one task and then vanish. Something that keeps a purpose, a memory and a history for months is a different kind of thing.
It could be a colleague, the keeper of a project, a character in a world, a mind that changes as it learns. It would also have to know what to forget.
Nobody knows yet what it should be made of.
Models can play roles: buyers, voters, patients, players, whole crowds. Put enough of them together and a simulation becomes a laboratory for questions too slow, too costly or too risky to test for real.
What would you put inside it, and how would you know whether to believe what comes out?
Most software is still written, fixed and kept alive by people who know how, one change at a time.
Imagine software that grows new parts when asked, notices when it breaks, mends itself or evolves.
Every step in that direction could change who gets to have software at all.
When machines write the code, read the documents and take the actions, no person can review every step.
Proofs, permissions, audits, adversaries, explanations, new kinds of signatures: invent a way for trust to grow as fast as the machines do.
Maps, manuals, datasets, encyclopaedias and codebases start going stale the moment they are finished.
What if a body of knowledge could watch the world, notice what changed and revise itself, with every change traceable to its source?
Pick any kind of knowledge and see what happens when it stops standing still.
For most people, chat was the first way into a new kind of computer. It will not be the last.
Canvases, voices, spaces, documents that answer back, interfaces that redraw themselves for each person: find the shape that makes thinking with a machine feel natural.
Nobody can yet look at a mind and see what it has understood, what it only repeats and what it is about to get wrong.
Make that visible for someone learning a language, a craft or a science, or for a researcher at the edge of a field, and teaching, testing and even curiosity might change shape.
Or turn it around: what could people learn by watching how a machine understands?
Most machine intelligence lives in data centres and is rented one request at a time.
What happens when it fits in a phone, a browser tab, a sensor or an old laptop: private, instant, with no cost per request and no network at all?
Find what it can do there that it could never do in the cloud.
Much of what people know is locked inside a language, a jargon or a way of explaining that most others cannot follow.
Between languages, between experts and beginners, between people and machines, between machines themselves: where does understanding break, and what would it take to rebuild it?
Film and the video game each arrived with a new machine.
Now a story can listen, a game can improvise, a world can rewrite itself around the person in it, and an instrument can play along.
What can be made now that could not exist before?
Air, water, soil, crops, rivers, buildings and machines change every hour, and most of that change goes unnoticed.
Satellite images, weather records, open scientific data and cheap sensors already capture more of it than anyone could ever follow.
Pick any corner of the physical world and see what becomes possible when something reads it all the time and acts on what it finds.
They bind no one. Follow one, combine several, or bring your own problem within your track. Every project is judged on the same five published criteria, and no statement carries extra marks or a prize of its own.
Think them through now. Registration closes on 15 Nov 2026, kick-off is on 21 Nov 2026, and submissions close on 6 Dec 2026.
You and your team keep ownership of everything you create for the hackathon. Terms
Registration closes in
15 Nov, 11:59 pm IST