All games
World difficulty

The standard world. Changing it restarts the whole game with a newly trained assistant.

Decision Lab

Mission Control

Launch rovers. Weigh the odds. Keep the operation alive.

A prediction is not yet a decision. Every mission asks the same question: given what the assistant thinks, what is actually worth doing — and what does it cost you?

Capital
—

Campaign

1

A prediction is not a decision

Recommendation only

Best score
—
Start
2

Can you trust the number?

Raw model score vs calibrated probability

Best score
—
Start
3

How uncertain is the number?

Same probability, different certainty

Best score
—
Start
4

WHY? What is it resting on?

Same odds, different evidence

Best score
—
Start
5

What could we change?

Model what-ifs vs real actions

Best score
—
Start
6

Better is not always more certain

Ensured alternatives

Best score
—
Start
7

Is the explanation plausible?

Guarded explanations

Best score
—
Start
8

Context changes uncertainty

Plains vs canyons

Best score
—
Start
9

Know when not to automate

Automation, reject and scarce specialists

Best score
—
Start
10

Build your operation

Open sandbox

Best score
—
Start

Installed upgrades

Upgrades are bought inside the levels that use them. Some levels can't run until you install what they need.