If you are integrating Persona into a real product, this is the core model to understand first.
The simple structure
The product is organized around three concepts:
- space
- app
- persona
The practical shape is:
space -> app -> persona
That is the cleanest way to think about the integration, even though the most important authority boundary is the space itself.
What a space is
A space is the top-level working environment for a deployment.
It is the main boundary for:
- tenancy
- runtime authority
- built-in capability surfaces
- token scope
- Persona residency
If you are integrating Persona for a customer, team, or world, the first question is usually: which space does this belong to?
What an app is
An app is the product-facing integration identity that lives inside a space.
Use an app when you want to connect Persona to:
- your product UI
- a specific workflow
- one integration surface
- one embedded experience
One space can host multiple apps. An app belongs to exactly one space.
That matters because apps help you separate product surfaces without losing the shared runtime context of the space they belong to.
Apps and external integrations usually reach Persona through runtime tokens scoped to that same space, which is why the app boundary and the space boundary need to stay aligned.
What a persona is
A Persona is the persistent AI runtime itself.
That is the entity that:
- remembers
- plans
- acts
- keeps continuity over time
A Persona lives in exactly one space at a time.
That is an important product rule. Personas are not supposed to drift between unrelated spaces, because that would break the continuity and authority model.
How the three fit together
The usual pattern looks like this:
- create or choose a space
- register one or more apps in that space
- create one or more Personas that live in that same space
- expose the Personas through the app surfaces you need
In other words:
- the space defines the environment
- the app defines the integration surface
- the Persona defines the persistent intelligence
Why this matters to integrators
Without this model, integrations become messy very quickly.
Teams start asking:
- where does this Persona actually live?
- which customer or world owns it?
- which app is talking to it?
- which token should be allowed to act on its behalf?
The space / app / persona model answers those questions before they turn into product confusion.
Design advice
If you are starting a first deployment:
- keep one clear space per customer, team, or world
- keep apps focused on specific surfaces
- keep Persona identity stable inside the owning space
That gives you the cleanest path to a system that feels coherent instead of improvised.