lucid tc | twingraph
Solutions · Gemini Enterprise + TwinGraph

Distribute Contextual Intelligence Securely Across Your Enterprise.

TwinGraph connects Gemini Enterprise to your operations, enabling real-time, bidirectional interaction with the systems that run your business.

Topology
Without TwinGraphyour operations — relationships not modeledMCPMCPMCPMCPGemini Enterpriseagent appADK AgentAgent RuntimeERPSAP / OracleIoT / MQTTsensor streamPLMproduct dataMESshop floorad hoc connectionMCP connectionAd hoc and MCP connections. Still no operations model.
With TwinGraphGemini Enterpriseagent appADK AgentAgent RuntimeMCP ServerCloud RunTwinGraph Serveryour operations · source of truthERPSAP / OracleIoT / MQTTsensor streamPLMproduct dataMESshop floor+ your systemany integrationone live graph. add a node → Gemini sees it.
The problem

Your AI sees the parts but not the machine. It can answer from yesterday's snapshot, but a plant manager needs to know how a sensor alert on Line 3 impacts the global supply chain today. When decisions depend on context that spans systems, a model that only reads documents will always be a step behind.

The TwinGraph angle

TwinGraph is the authoritative nervous system for your enterprise: a live graph that unifies ERP, IoT, and PLM into one queryable, actionable model. With our MCP Server on Cloud Run and the Google ADK, TwinGraph-powered agents deploy to the Gemini Enterprise Agent Runtime (formerly Vertex AI Agent Engine) and reach every authorized user through Gemini Enterprise Apps. Add a system to the graph and every Gemini Enterprise App sees it instantly.

In practice

From question to action, in seconds.

Why is Line 3 running 12% under target this shift?
Plant manager, in Gemini Enterprise
  1. 01 · Trace

    The agent walks the live graph: a torque sensor on Press 7 has been throwing warnings for 90 minutes, and an upstream PO from Supplier A is delayed.

  2. 02 · Reason

    It cross-checks the maintenance schedule and sees tonight's window collides with the recovery plan. It surfaces the conflict in plain language.

  3. 03 · Act

    With one approval, the agent reschedules maintenance to Friday, shifts two SKUs to Line 5, and writes both changes back to the MES through TwinGraph.

  4. 04 · Audit

    Every read and write is scoped to the manager's IAM role and logged, so security and ops can trace exactly what the agent saw and did.

How it works

A few steps. Real infrastructure.

01

Deploy TwinGraph

Unify ERP, IoT, and PLM data into a single live graph, hosted on a secure TwinGraph Server.

02

Host the MCP bridge

Run the TwinGraph MCP Server on Google Cloud Run as a production bridge between agents and the graph.

03

Build with Google ADK

Compose an agent with Google's Agent Development Kit, attach the TwinGraph MCP toolset, and deploy to the Gemini Enterprise Agent Runtime.

04

Distribute via Gemini

Publish the agent as a Gemini Enterprise App, gated by Google Cloud IAM so the right people can query and act safely.

Runtime architecture
chatchatinvokeinvoketool calltool callread & actread & actEnterprise Usernatural languageGemini Enterpriseagent appADK AgentAgent RuntimeMCP ServerCloud RunTwinGraphoperational graphExternalPlatformSource of truth
Why TwinGraph
  • 01

    Decisions in seconds, not hours: leaders ask in natural language and get answers grounded in what's true right now, across every connected system.

  • 02

    Act, don't just read: agents can adjust operational parameters in place, turning insight into action without leaving the chat.

  • 03

    Enterprise-grade by default: every query and write inherits your existing Google Cloud IAM, so authority follows the user, not the model.

See this running on your infrastructure.