Skip to content

Core concepts

Orbismo has two sides: your AI and the portal. Understanding the split makes everything else click.

Your AI writes. Every change to your world’s contents happens through conversation with the AI you’ve connected. Adding a character, updating a detail, recording an event, deleting something outdated: you describe it, and your AI creates and connects the entities, relationships, and lore for you. The world’s structure is edited in the portal instead, under Building Blocks.

“Add my sister Sarah, she lives in Denver.”

“Sarah moved to Portland last month.”

“Delete the old draft of chapter three.”

The portal reads. The web app at app.orbismo.com is where you and your collaborators browse, search, and explore the world your AI has built. It’s also where you manage collaborators, keys, and billing. When you spot something in the portal that needs changing, tell your AI.

This split is what keeps your world consistent. Because every change flows through the AI, new information gets connected to what’s already there instead of piling up in disconnected notes. That’s how Sarah stays one person with one history, whether she was mentioned in a chat last week or six months ago.


Orbismo has a small vocabulary, and these few terms carry you through the whole product.

The top-level container. A world is one self-contained universe — its own entities, lore, collaborators, and subscription. You can own several worlds, and each is independent of the others — like separate campaign binders, casebooks, or project files.

The building block of a world. An entity is any element you want to track — a person, a place, an event, a group. Entities are the nodes of your world’s graph, and each one can carry custom properties and lore. See Entities and relationships.

The category an entity belongs to — Person, Place, Event, and so on. Every world starts with a default set of entity types and can be extended with your own.

A connection between two entities — “is a member of”, “lives at”, “led to”. Relationships are the edges of the graph; they’re what let you ask connected questions like “everyone in this group who attended that event.” See Entities and relationships.

Long-form prose attached to an entity — its history, description, and detail. Lore is indexed for semantic search, so you can find things by meaning, not just by exact wording. See Lore.

A label from a shared vocabulary you apply across entities (for example, “coastal” or “noble house”). Tags keep your world consistent by preventing near-duplicates like “swordsmanship” vs. “sword fighting.”

Every world starts from a general-purpose default schema, a set of entity types (Person, Place, Event, Group, Project, Interest, Item, Reference, Rule) and the relationship types that connect them. You pick a starting template in the creation wizard, and the general-purpose My World template seeds exactly this schema. You then adapt it to whatever you’re building, changing the schema itself under Building Blocks and steering how it’s used with world instructions. See the Data model.

Other people you invite into a world. Each collaborator has a role:

  • Owner — the person who created the world. Full control, including billing, invitations, and roles. Every world has exactly one owner.
  • Editor — can read and write entities, lore, and relationships, but can’t manage billing or invite people.
  • Viewer — read-only access.

See Collaborators and permissions.

Every world is reachable over the Model Context Protocol (MCP), so AI tools can read and edit it alongside your team. Access is governed by you: sign in through the connector, or issue an API key — either way, read-only connections can query and read-write connections can change the canon. See Connect an AI tool.