Solutions · Gemini Enterprise + TwinGraph

Distribute Contextual Intelligence Securely Across Your Enterprise.

Move beyond static documents. TwinGraph connects Gemini Enterprise to your living operational graph—enabling real-time, bidirectional interaction with your real-world systems via Google's Agent Development Kit (ADK) and Vertex AI.

Talk to an engineer

Topology

The problem

Current AI implementations struggle with 'data silos'—they see the parts but not the machine. Static RAG can't keep up with operational systems. Gemini can answer from yesterday's snapshot, but enterprise leaders need to know how a specific sensor alert on the factory floor impacts the global supply chain today.

The TwinGraph angle

TwinGraph acts as the authoritative 'nervous system' for your enterprise. By deploying our secure MCP Server to Cloud Run and utilizing Google's Agent Development Kit (ADK), you can deploy TwinGraph-powered agents directly to the Vertex AI Agent Engine. This enables AI workflows grounded in real operations to be distributed throughout an organization securely via Gemini Enterprise Apps.

How it works

A few steps. Real infrastructure.

Deploy the Digital Twin

Model disparate data (ERP, IoT, PLM) into a unified graph schema and deploy securely to TwinGraph Server via gRPC.

Host the MCP Bridge

Deploy the TwinGraph MCP Server to Google Cloud Run to create a production-ready bridge.

Build with Google ADK

Create an AI agent using Google's Agent Development Kit, attaching the TwinGraph MCP toolset, and deploy to Vertex AI.

Distribute securely via Gemini

Import the agent into a Gemini Enterprise App. Apply Google Cloud IAM permissions so authorized leaders can safely query and modify operational parameters.

Capabilities used

Google Agent Development Kit (ADK)Vertex AI Agent EngineContextual RAGBidirectional Action ExecutionEnterprise IAM Integration

Example code

Real Building the TwinGraph Agent via Google ADK.

python · gemini-enterprise.pylive
# Create an ADK agent powered by Gemini with Access to the twingraph_mcp_server toolset
from google.cloud.aiplatform import reasoning_engines

root_agent = reasoning_engines.Agent(
    name="twingraph_agent_123",
    description="Agent that can call MCP-exposed TwinGraph tools.",
    model='gemini-2.5-flash',
    tools=[twingraph_mcp_toolset, get_time],  # Attach the MCP toolset
    instruction=(
        "You are an intelligent agent powered by a TwinGraph Digital Twin. "
        "You can answer questions and perform secure actions related to all of the data sources, "
        "systems, and semantic information embedded in the TwinGraph Digital Twin. "
        "You can call the available MCP tools when needed. "
        "Validate input parameters against each tool's schema."
    ),
)

# Deploy to Vertex AI Agent Engine
reasoning_engines.ReasoningEngine.create(
    root_agent,
    display_name="TwinGraph Operations Agent"
)

Why TwinGraph

01
Faster Decision Making: Unified Graph Schema normalizes disparate data into a relational web, reducing time-to-insight from hours to seconds.
02
Operational Agility: Bidirectional integration allows real-time adjustments to physical processes via natural language.
03
Risk Mitigation: Inherits enterprise-grade IAM permissions, ensuring high-level twin modifications are only made by authorized personnel.

Contextual RAG

Retrieval that understands the relationship map and interconnected systems.

Orchestrated Operations

Your operational systems, modeled as a living graph.

See this running on your infrastructure.

Request Demo