Founders
Find a problem worth a company: big, open, and not already crowded.
The map of everything left to solve.
A research lab mapping the world's open problems: what each one is worth, who is working on it, and what gets solved next.
Want to see the map as it grows? Leave your name, and we'll be in touch.
Explore
Illustration
01The bottleneck
AI gave us speed. Not direction.
A small team with AI can now build in weeks what used to take years. But the fastest car in the world doesn't help if you don't know where you're going. You stop, you ask, you drive in circles, only faster.
Same car. Same speed. One of them has a map.
The bottleneck isn't intelligence anymore. It's knowing what to work on.
Without a map, cheap building makes it worse. Everyone races to the same easy ideas and copies each other within weeks, while the hard problems, the ones that matter, stay unsolved.
~7 moHow often the length of tasks AI can finish on its own has doubled, since 2019.METR, 2025 ↗
18×the researchers it now takes to double a chip's power, compared with the early 1970s.Bloom et al., AER 2020 ↗
90%+less disruptive than in 1945, by one measure: new research rarely opens a new direction.Park et al., Nature 2023 ↗
02The map
Work is connected. So it can be mapped.
Every problem sits next to others: the ones it depends on, the ones it unlocks, and the ones someone else is already solving. Connect them, from the whole economy down to a single task, and you get a map. Think of it as Google Maps for the world's work.
US economyHealth careBillingPrior authorization
1.0×- 01 · The economy
Start with everything.
The US economy, about $33 trillion a year, split into industries and sized by the value each one adds.
- 02 · An industry
Open one industry.
Health care and social assistance adds 7.5% of it. Inside is the work that keeps it running: care, diagnosis, records, billing.
- 03 · Problems
Zoom in again, and you reach problems.
Each one is a place on the map: solved, being solved, or still open. The lines show what each problem needs before it can fall.
- 04 · One problem
Look before you spend a year on it.
Open a problem and see what it's worth, who needs it, who is closest, and what it unlocks next.
Industry sizes: value added by industry as a share of US GDP, second quarter of 2026 (BEA, via FRED). US GDP: third quarter of 2026, annual rate. Everything inside an industry, and the problem card, is an illustration.
03Forecasts
Add time, and the map becomes a forecast.
Problems aren't solved at random. They fall when the problems beneath them fall, when the tools arrive and when the right people get close. Google Maps tells you when you'll arrive. Our map tells you which problems fall next, when, and who gets there first.
Billing & insurance · 12-month forecastNow
% = chance it's solved within 12 monthsIllustration
04The record
A map you can check.
Anyone can draw a map. Ours has to be right, and you should be able to see when it isn't. So we test it in the open.
- First
On the past
The map sees only what was known at the end of 2024 and forecasts 2025. We score it against experts and simple baselines.
- Then
Live
New forecasts go out with a date, before anyone knows the answer.
- Always
In public
Every hit and every miss stays in the record. The record is how you know when to trust the map.
No.ForecastChancePublishedResolvesResult
0001The five largest US health insurers approve most routine prior authorizations automatically.61%Nov 2026Nov 2027Open
0002Most routine outpatient visits at large US hospital systems are coded with no human review.74%Nov 2026Nov 2027Open
0003A robot that takes laundry from the basket to folded goes on sale to US homes.22%Nov 2026Nov 2027Open
Example entriesThe first real forecasts are published when the benchmark opens.
05Who it's for
For anyone deciding what to work on.
Companies
See where your market is going, and what to build before someone else does.
Investors
See what's about to move, and who is closest to moving it.
Researchers
Find the questions that unlock the most, and see who's already on them.
We use it first. The map picks what we build next, so we feel its mistakes before anyone else does.
06Principles
How we draw the map.
Legend5 symbols
-
Value, measured.
Every problem carries a number: what solving it is worth, and to whom.
-
Every forecast, dated.
We say when, in advance. A forecast without a date can't be wrong, so it can't be trusted.
-
Connected, not listed.
A problem only makes sense next to what it needs and what it unlocks.
-
Signals, not opinions.
We follow what people do (papers, code, patents, hiring and funding), not what they say they'll do.
-
People choose the destination.
The map shows the routes. People decide where to go.
07Contact
What should the world work on next?
That's the question behind every other question. If you want to see the map as it grows, leave your name, and we'll be in touch.
Or write to [email protected]. Investors and partners: the same address.