xoul.ai

A Xoul is the character the AI plays, and you write its name, personality, greeting, and rules so the model can use that text as a script rather than as code. Up to 8 Xouls can share one chat, which is what makes ensemble scenes possible. Paid tiers raise the character limit, so a complicated character with a long backstory has more room to breathe without getting trimmed. More room, deeper characters.

Scenarios set the premise that opens a story. They come in three shapes: no embedded Xouls, one locked-in Xoul, or 2 to 8 Xouls locked into a group chat. A Scenario can be as short as a single line of plot or as long as a full paragraph describing the setting and premise, which means you can seed a quick scene or a richly detailed world depending on what you need. Creators often park a narrator plus a placeholder Xoul so you can swap in whoever you want. Flexible by design. No forced cast.

Give a Scenario an Objective and hit Generate Meters, and the platform builds four progress trackers from your goal. Meters show up in the chat's side panel and can nudge the story as they move. Light game layer. When you want a long chat to keep a direction instead of drifting, those trackers do more work than most people expect.

A Persona tells the AI your role in the story without the model playing it back, and Memories hold current events and details the model should remember so the character stops forgetting what happened ten messages ago. Together they fix the biggest roleplay headache: characters forgetting established facts. Better memory is a concrete perk of paid tiers.

Lorebooks store background the AI pulls in when relevant, like locations, factions, rules of a magic system, or recurring side characters that keep showing up in a long story. You keep the entry, and the platform feeds it into context as needed, so you don't have to repeat yourself every few messages. Set it once, then forget it.

Each chat lets you pick a model, and unlimited access to all models starts at the Purple tier, which is worth knowing if you care about tone more than raw speed. There's also a Custom Engine (Proxy) option for users who want to route replies through their own setup. Swapping models is the fastest way to test a scene.

You can adjust how long and how stylized replies are, which matters a lot in roleplay because short, punchy responses suit fast dialogue while longer ones help descriptive scenes build atmosphere. Small setting. Big difference.