Your Code. Your Models. Your Infrastructure. Always.

Governments can compel AI companies to retain data and change their privacy practices. Fabric moves with you: reconfigure to on-premise at any time, bring all your work with you, and switch models with a configuration change.

Fabric Enterprise is an AI coding IDE built for sovereign deployment. It can be reconfigured to run entirely on-premise at any time — all conversation histories, context, and work come with you. Switch between hosted models and self-hosted open-weight models. No vendor lock-in. Available for macOS, Windows, and Linux.

Built for Security-First Organizations

Every architectural decision in Fabric prioritizes data sovereignty and compliance.

On-Premise in One Move

Reconfigure Fabric to run entirely on-premise at any time. All conversation histories, context, and work migrate with you. No data left behind, no vendor lock-in. Your work is yours — architecturally, not just contractually.

Seamless Model Switching

Switch between hosted frontier models and self-hosted open-weight models with a configuration change. When circumstances shift, your development tools don't have to.

Sovereignty by Architecture

License terms can be overridden by government mandate. Fabric's sovereignty is architectural — your data physically cannot leave your environment in on-premise mode. No policy change, executive order, or acquisition can change that.

Your Models, Your Keys

Bring any model, self-hosted or through a managed API with your own keys. Local models can trade some peak performance for sovereignty. That trade-off is yours to make, not your vendor's.

Encrypted in Transit and at Rest

Data is encrypted in transit and at rest, and API keys are additionally encrypted at the application layer. In on-premise deployments, encryption follows your own key-management policies.

Zero Data Retention on Inference

Fabric-hosted inference runs with zero data retention: prompts, code and responses are not stored, logged or used for training. In a world where AI companies are being compelled to retain user data, there is no inference record to compel. If you need a record, your enterprise organization can switch on a conversation archive for its own members: it stays off until you turn it on, every member is notified and must acknowledge it before anything is copied (unless you turn the notice off after confirming staff were told another way), and you set how long it is kept. An archive you switch on is a stored record, and a stored record can be compelled; leave it off and there is still no inference record to compel.

Built for Regulated Environments

Whether you work under GDPR, Canadian privacy law, or classified-handling rules, you can deploy Fabric where your obligations require.

European Union & GDPR

GDPR restricts transfers of personal data outside the EU. Fabric can be deployed on-premise within your EU infrastructure, so your code and prompts stay inside your environment.

Canada

Fabric is built by a Canadian company. Hosted inference runs only through providers bound by zero-data-retention agreements. For enterprise deployments, we can provision dedicated GPU capacity hosted in Canada, so inference stays in-country. On-premise, air-gapped and local deployments give you full control over residency.

Defense & Classified Environments

Fabric can be deployed on-premise or fully air-gapped, with no external network calls. Because it runs inside your environment, workloads such as ITAR-controlled or Protected B data stay under your own authorization.

Residency Under Your Control

With an on-premise deployment, you choose where your code, prompts and AI context live, without relying on a vendor's data-center locations.

Enterprise Capabilities

Usage Analytics & Reporting

Dashboards showing AI usage by team, model, and project, so you can track consumption and allocate cost.

SSO & Identity Federation

Single sign-on with Google, Apple, and GitHub today. Enterprise SSO through your organization's identity provider is on our roadmap.

Role-Based Access ControlOn the roadmap

Granular control over who can access which models, view analytics, manage billing, and configure workspace rules.

Custom Fine-Tuned ModelsOn the roadmap

Models trained on your codebase, coding standards, and domain knowledge, so they write code the way your team does.

Ready to Deploy?

Talk to our team about on-premise and air-gapped deployment, residency, and terms for your organization.

Contact Sales

Enterprise FAQ

Can Fabric run completely air-gapped with no internet access?

Yes. Fabric can be reconfigured at any time to run entirely on-premise with no external network calls. Your conversation histories, context, and work come with you. Run self-hosted models locally and the IDE works fully offline.

What happens if I need to move off cloud AI providers?

That's exactly what Fabric is built for. Switch from cloud models to self-hosted alternatives with a single configuration change. Your work, context, and history are stored locally — nothing is held hostage in a vendor's cloud. You trade some peak performance for definitive sovereignty, and that trade-off is yours to make whenever you want.

Why can't I just rely on my AI provider's privacy policy?

Because privacy policies are license terms, and governments override license terms. OpenAI was ordered to retain conversations past their agreed expiration dates. AI companies face potential lawsuits, fines, or national security mandates that can change their data practices overnight. Fabric's sovereignty is architectural, not contractual — in on-premise mode, your data physically cannot leave your environment.

What models can I use on-premise?

Any model you can host and serve through an OpenAI-compatible API, such as open-weight models running on vLLM or Ollama. Self-hosted models can trade some peak performance against the largest hosted models, in exchange for full control.

How is Fabric different from Cursor or Copilot for enterprise?

Cursor and Copilot are cloud services. Fabric can be reconfigured to run fully on-premise at any time, brings all your work with it, and lets you switch models with a configuration change, so you are never tied to one vendor's cloud.

What's the deployment architecture?

Fabric can be deployed in your own environment, on-premise or in a private cloud, including fully air-gapped. Our team works with you on an architecture that fits your security and residency requirements.