Future PresentLabs

FPL Cloud / Preview

Run AI.
Keep authority.

Your models. Your infrastructure. Explicit boundaries for what agents can do.

Request AI access →

Models & provider connections

Bifrost is the inference gateway. Loki connects supported provider identities. We’ll confirm which provider, model and authentication path fits your workload; connecting an account does not imply unlimited usage or that every provider permits the same integrations.

Agents & tools

Edgerunner orchestrates supported agent runtimes. FPL SSO and the MCP Gateway provide identity and scoped tool access. Runtime, adapter and deployment support are confirmed during preview onboarding.

Govern the action, not just the prompt.

Identity and scoped grants define which tools an agent may use. Edgerunner coordinates execution; Mycelium and Unibus provide the infrastructure and capability fabric. DCP makes bounded decisions explicit rather than leaving control flow to a prompt.

Explore the Decision Catalog Protocol ↗

Open core. Optional operations.

Sovereignty means being able to operate without us. Core policy, distribution and node metrics belong in the open foundation. Our managed direction adds hosted authority, console visibility, audit, alerting and enterprise support—not a paywall around the core. We confirm available capabilities and deployment boundaries during onboarding.

Our open-source commitment →

Know what you’re paying for

FPL-billed inference tokens and GPU inference are prepaid. Other cloud services are billed monthly and require saved-card autopay. Provider-connected usage depends on the supported integration and provider terms; we’ll clarify attribution and charges before activation.

What happens next

Request access with your email and project details. We’ll confirm compatibility, availability and pricing. The current request form does not activate a paid service. Unified FPL SSO registration is being prepared; public self-serve activation is not live yet.

Looking for Verk, Transmog or Legion-of-BOM? Explore our engineering AI work →