Skip to content

Integrations

Forge Pool is designed to be embedded directly into real systems.

Forge Pool integrates once.

All surfaces share the same execution core.

Forge Pool does not provide separate systems for different users.

All integrations map to the same execution layer.

The difference is only in how execution is constructed and initiated.


Integration Surfaces

Forge Pool currently exposes three primary integration surfaces:

SurfacePrimary UserPurpose
StudioHumansCompose and explore execution systems
APISystemsIntegrate deterministic compute
MCPAI AgentsExecute and reason over uncertainty

1. Studio (Human Interface)

Studio provides a visual execution environment for building and operating probabilistic workflows.

Capabilities include:

  • visual composition of execution systems
  • workflow orchestration through blocks
  • debugging and execution tracing
  • output exploration and distribution analysis

→ See: Forge Studio


2. API (System Interface)

The Forge Pool API exposes the canonical execution contract used by the system itself.

Capabilities include:

  • deterministic execution
  • replayable workloads
  • execution inspection
  • distributed compute orchestration
  • auditable execution metadata

→ See: API


3. Agent Access (MCP)

Forge MCP exposes Forge Pool directly to compatible AI agents.

Agents can:

  • discover capabilities
  • inspect execution contracts
  • execute workloads
  • retrieve compact results
  • analyze distributions
  • replay deterministic executions

→ See: MCP


Design Principle

Forge Pool does not expose isolated product features.

Forge Pool exposes execution capability.

Every integration surface ultimately maps to the same execution system:

  • primitives
  • profiles
  • adapters
  • execution contracts
  • deterministic execution infrastructure

The difference is only:

  • who is initiating execution
  • and how execution is constructed

Distribution-First Execution

Traditional systems typically produce:

  • single outputs
  • static predictions
  • opaque execution paths

Forge Pool instead operates on:

  • probabilistic execution
  • scenario exploration
  • distribution surfaces
  • replayable deterministic workloads

This enables:

  • tail-risk analysis
  • uncertainty exploration
  • scenario comparison
  • auditable reasoning under uncertainty

Final Note

Forge Pool is not a collection of disconnected tools.

It is a unified execution substrate accessed through different interfaces.


Next:

MCP