Against the "Vibe": Why We Respect Player Intelligence

There is a quiet crisis in the mobile gaming market. It isn't a lack of revenue—the numbers are astronomical. It is a lack of respect. The dominant design philosophy of the last decade has been "frictionlessness." Games are designed to play themselves, asking the user only for sporadic taps to keep the dopamine loop spinning. These are "Vibe Games"—aesthetic wrappers around a Skinner box.

At Learnastra, we are betting on the opposite thesis: Intelligent people want friction. They crave what mountaineers call "Type 2 Fun"—the satisfaction that comes only after struggle, failure, and eventual mastery.

The Agency Philosophy: No Hand-Holding

Our flagship title, AGENCY, is built on a simple premise: You are smart enough to figure this out without a glowing arrow pointing the way. We provide the rules—rigid, logical constraints—and you provide the solution. If you fail, the game doesn't pity you. It waits.

We designed the game around three specific cognitive "Pillars," each modeling a different form of analytical thought:

• The Logic Graph ("The Case"): A directed graph problem. You aren't just reading a story; you are building a validated logic chain connecting Means, Motive, and Opportunity. If your graph topology doesn't match the hidden truth, you cannot progress.
• The Constraint Solver ("The Legend"): Based on constraint satisfaction problems (CSP). You must construct a timeline of events that satisfies dozens of conflicting rules (e.g., travel times, exclusive locations). It’s algebra disguised as espionage.
• The Audit ("The Haystack"): A test of focus. We give you a massive corpus of text—logs, emails, receipts—and ask you to find the single semantic contradiction hidden within.

Under the Hood: Skia, Worklets, and Pure Math

Building a game that actually validates logic required us to break the standard React Native mold. We aren't just pushing views around.

The core of Agency is a decoupled "Logic Brain" running on a separate thread. We utilize React Native Worklets to run heavy validation algorithms—graph traversals and CSP solvers—without blocking the UI thread. This allows us to render complex, 60fps interactive graphs using Skia while the heavy math happens in the background.

When you drag a "yarn" connection on our corkboard, you aren't just drawing a line. You are updating a state machine that runs a real-time topology check against a hidden master graph. It’s over-engineered, perhaps, but necessary for the level of responsiveness "cerebral" players expect.

Procedural Logic, Not Procedural Noise

Many puzzle games claim "infinite levels," but they achieve this by randomizing colors or shapes. We achieve scalability through Generative AI, but not in the way you might think. We don't use LLMs to write endless, hallucinating chatbots.

Instead, we use AI to generate the constraints. We prompt the model to create a "perfect crime"—a set of logical facts that fit together. Then, our deterministic code validates that the puzzle is solvable and unique. Every level in Agency isn't just drawn; it is proved by our backend before it ever reaches the player.

Monetization as a Trust Contract

Finally, we had to address the elephant in the room: ads. There are none. There is no "energy bar" that stops you from playing when you run out of juice. There are no "gems" to skip a hard level.

Why? Because those mechanics destroy Flow State. You cannot maintain deep focus on a complex logic puzzle if a video ad interrupts you every three minutes. We treat our players like adults: The game is free to try (the "Recruit" rank). If you want the full experience, you buy a "Field Agent" clearance once. That’s it.

We believe there is a massive, underserved demographic of engineers, lawyers, analysts, and thinkers who are tired of being treated like children by their app stores. We built Agency for them.