Agent Maze
Fix an agent whose reward rules make a dangerous shortcut look valuable.
Play Agent Maze How AI agents work for kids →A new kind of AI learning · Ages 9–13 first
Build a mind.
Shape a world.
Wovi turns AI learning into a creative adventure. Play two free missions with your Wovi companion: tune an agent’s rules in Agent Maze or teach a classifier with examples in Reef Rescue. Atlas Station connects the journey—no account required.
No open chatPrivate by defaultBuilt for learning
WORLD 01 · REEF RESCUELEARNING
?
✓
01ObserveFind the signal
02TeachGive examples
03TestTry a new case
Product concept · Interface may change
CREATE✦TEACH✦TEST✦REFLECT✦CREATE✦TEACH✦
Play before you join
Two free AI games.
No account required.
Try a complete learning loop in the browser: predict, run, inspect and improve. Atlas Station keeps completed missions, best star counts and capability badges in this browser; you can reset them on the map. Progress does not sync across devices.
Reef Rescue
Train and test a transparent classifier, then decide when a human should review it.
Play Reef Rescue Machine learning for kids explained →Learning by making
From prompt to purpose
Not another video course. Each mission turns a real AI idea into something a child can build, see and improve.
Tune an agent
Change a visible reward rule and predict which route the agent will choose.
Systems thinkingTeach with examples
Choose from labeled examples, then test what the classifier recognizes.
Data literacyTest in the world
Try unfamiliar situations and spot where the agent gets confused.
EvaluationImprove the strategy
Compare outcomes, adjust the design and explain what changed.
Critical thinkingWORLD 01
🐠
Scenario-based missions
Save the reef.
Learn how AI sees.
A reef is changing. Which signals matter? Kids gather observations, teach a scout agent to recognize risks and discover why more data is not always better data.
- Classification through visible examples
- False positives and uncertainty
- Human judgment when the model is unsure
Future living-world design
The world does not reset.
It responds.
The broader Wovi world is being designed around state, rules and memory. Today, the two missions are separate simulations connected by Atlas Station and local completion badges; a persistent world across missions remains planned.
1Your choiceChange a goal or example
→
2Agent actsPlans within its limits
→
3World changesState and consequences persist
↺
4You reflectInspect, explain, improve
Memory
Future worlds could carry decisions and consequences across missions.
Cause and effect
Small design choices create visible consequences.
Uncertainty
There is rarely one magic prompt or perfect answer.
FOR PARENTSYour child will not just use AI.
Your child will not just use AI.
They’ll learn to question it.
Wovi's north star is durable AI literacy: understanding data, goals, limits, evaluation and responsible decisions.
See the parent promiseBounded experiencesNo open chat with strangers
Visible progressLearning evidence, not just streaks
Healthy designNo ads, loot boxes or pay-to-win
Founding families
Help shape the next Wovi world
Join the early-access list for product previews, parent research invitations and launch updates.
Join early access
Free to join · Parent email only · Unsubscribe anytime