A quiet glass sanctuary on a basalt island above the clouds.

Private AI · Early access

A place of its own.
An AI of your own.

Your personal cloud. Built around your boundaries.

We’re exploring a private home for Prism: your own hosted environment for your context, agents and open models. Tell us how you’d want it to work.

Discuss early access Explore the idea

Inside the experience

Your world. A home for its intelligence.

A personal cloud brings the workspace and model into an environment set up for you. The important choice is what stays inside it—and what you allow to connect outside.

Your devicesWhere you work

Your personal cloudProposed environment

Your context

Selected files & memory

Prism

Your agents & workspace

Your model

Hosted in this environment

External services you choose

Connected apps · Optional cloud model APIs

Connected services receive the data needed for their actions. Cloud model APIs process prompts outside your personal environment.

Proposed setup. Hosting, models and connections are agreed for each pilot.

The idea

More personal. By design.

For people who want a closer relationship with their AI—and a clearer picture of where their work goes.

A space set up for you

Explore a separate environment for your workspace, selected context and AI models.

More say in the model

Choose the balance of capability, speed and cost that fits your actual work.

Clearer data boundaries

Agree what runs inside, which services can connect, and who can access the environment.

Understand your options

Where should your AI live?

Hosting changes who looks after the system and where information is processed. These are the choices to discuss—not three plans you can buy today.

The familiar starting point

Managed cloud

Use the hosted app and its available model services.

Good for: Getting started with less setup
Keep in mind: The app and model providers process the data needed to run the service.

What we’re exploring

Personal cloud

A private hosted environment for your workspace and model.

Good for: More control without hardware at home
Keep in mind: Hosting costs continue. The infrastructure provider remains part of the setup.

A different ownership model

Your own hardware

Run compatible software and models on a machine you own.

Good for: Physical control of your compute
Keep in mind: You maintain the machine. A fully local Prism package is not being offered here.

Connected, deliberately

Keep the useful connections.

Private hosting still needs to work with your world. Choose the accounts and tools you need, then review how each connection handles data. Compatibility would be confirmed for your environment.

Explore integrations →

Good to know

A few things, answered.

Can I get this today?

We’re exploring private hosting with interested users. Contact the team to discuss a pilot. Availability, deployment details, support and pricing need to be agreed before you start.

Is a personal cloud the same as owning a computer?

A personal cloud is a hosted environment set up for you. The infrastructure belongs to a hosting provider, which remains part of the trust boundary. Running on your own hardware gives you physical control, with responsibility for power, updates, backups and remote access.

Does all my data stay inside?

That depends on the agreed setup. A model hosted inside the environment can process prompts there. Connected apps still communicate with their providers, and any external model API receives the prompts sent to it. Backups, logs and support access also need clear boundaries.

Which models would it run?

Open models are the starting point for the idea. The right choice depends on your tasks, model licence, available memory and response speed. We would evaluate quality on your use cases before recommending a configuration.

Will it keep working while my laptop is closed?

A hosted environment can stay online independently of your laptop. Which Prism agents, schedules and connections can run there is part of the pilot evaluation. Continuous operation also depends on the host, network and connected services.

What would it cost?

Pricing is not announced. Compute, storage, backups, model size and support all affect the cost. Always-on GPU hosting has ongoing running costs; we would agree a budget and scope before any pilot.

How it comes together

Tell us your needs

Share your use cases, preferred region, data boundaries and budget.

Shape the setup

Discuss hosting, model quality, connections and who looks after the environment.

Evaluate together

Agree scope, availability and cost before any pilot begins.

Inside a warm personal workspace overlooking mountains at dawn.

Make room for
your own AI.

Discuss early access

A place of its own.An AI of your own.