A watercolour game scene: a hooded adventurer walks a torchlit cave path towards a glowing door, past a treasure chest.

Framework for Voice Controlled Games

Players talk. Your game listens.

Hears everything, acts on what matters.

AI filtering cuts the chatter. Only real requests reach your game.

Six things players said, as one band of sound. "Give me a giant sword." and "Start a thunderstorm." are direct asks, and the game acts on them. "spiders freak me out" is context, remembered as "Scared of spiders". The chatter, like "brb, pizza’s here", is dropped.

Games react to your voice.

Customize the voice framework for your game.

Your game decides what it can do, watch, remember and spend.

  1. 1Your actions

    Expose your game’s actions as MCP tools. Only the ones you allow ever run.

  2. 2Your triggers

    Pick the game events it watches and what it listens for.

  3. 3Reusable actions

    What it builds is saved and reused, so a repeat ask comes back fast.

  4. 4Your memory

    Decide what’s kept for each player and for the whole world.

  5. 5Your models and budget

    Pick the model for each job, and cap what each player costs.

A player's line goes through the AI filter into the voice framework, and the result reaches your game. Five modules plug into the framework: your actions, your triggers, reusable actions, your memory, and your models and budget.

Remembers every player.

Three layers, loaded into every conversation.

Player memoryWhat each player likes, says and reacts to, kept for later.
World memoryWhat’s true for the whole world, shared by everyone in it.
Live contextA running summary of every conversation, so the whole game stays in sync.

Yours to delete.

Give your game ears.

Join the waitlist