FactVerse AI Agent

FactVerse AI Agent connects FactVerse digital twins, operational data, simulation services, knowledge, and governed execution paths. Implementation teams use it to build agent workflows for facilities, equipment, maintenance, simulation-ready assets, and industrial operations.

Use this page to choose the right document for the task in front of you.

Platform model

FactVerse AI Agent is organized around configured agents, governed tool access, data readiness, review gates, and audit records. The platform model is broader than an MCP endpoint: MCP exposes tools to approved clients, while the Agent Platform decides which agents exist for a tenant, which tools each agent may use, and how higher-risk actions are reviewed.

Start with Agent Platform Overview when explaining the overall runtime. Use Agent Hub when planning the product entry point for tenant-specific agents.

Implementation flow

Choose your path

I need to	Start with	Then use
Understand the governed Agent runtime	Agent Platform Overview	Architecture
Plan how users find configured agents	Agent Hub	Workflow Guides
Configure, change, or retire an agent	Agent Lifecycle and Configuration	Workflow Run Record
Define risk levels and review gates	Risk Governance and Safety	Access and Scope Planning
Confirm available tools at runtime	Runtime Tool Discovery	MCP Tool Reference
Design tool call outcomes	Tool Execution Governance	Workflow Run Record
Connect an AI client for the first time	Getting Started	MCP Integration Guide
Plan endpoints, API keys, scopes, and approvals	Access and Scope Planning	MCP Scope Matrix
Check whether the workflow has enough usable data	Data Readiness	DFS overview
Build a facility operations workflow	Facility Operations Workflow Guide	Facility Operations use case
Build a predictive maintenance workflow	Predictive Maintenance Workflow Guide	Predictive Maintenance use case
Build a Physical AI workflow	Physical AI Workflow Guide	Physical AI use case
Review request and output patterns	Examples	Workflow Run Record
Diagnose a failed client, scope, data, or write action	Troubleshooting	MCP Errors and Audit
Move a workflow from pilot to regular use	Validation and Handover	Workflow Guides
Look up available tools and scopes	Tool Reference	Scope Reference

Operating model

A production workflow should follow this order:

1. Define the agent or workflow owner.
2. Verify tenant, site, asset, equipment, scene, and source-system boundaries.
3. Check source freshness, signal quality, document context, scene version, and review ownership.
4. Confirm runtime tool discovery for the configured agent or client.
5. Add compute or simulation steps only after the inputs and assumptions are visible.
6. Put controlled actions behind confirmation, human approval, and audit records.
7. Capture final evidence, reviewer decisions, operator feedback, and accepted corrections.

Use Core Concepts for the vocabulary behind this operating model. Use Architecture and Capability Map when planning a larger deployment.

Product context

Product area	Role in an Agent workflow
FactVerse Platform	Tenant, asset, permission, and product context.
Data Fusion Services	Source connection, mapping, quality checks, governed datasets, fusion, review, and BI-ready data.
FactVerse Designer	Digital twin scenes, simulation-ready assets, layout planning, and scenario preparation.
DataMesh Inspector	Runtime visualization, alarms, work orders, inspection records, and facility operations context.
MCP endpoints	Governed tool access for AI clients and enterprise agent runtimes.

Expected implementation output

For each Agent workflow, keep these artifacts visible:

Artifact	Purpose
Workflow contract	Defines role, boundary, inputs, output, reviewer, and approval path.
Access package	Lists MCP endpoint, scopes, key owner, rotation rule, and client runtime.
Data readiness note	Records source systems, mappings, freshness, quality issues, and missing evidence.
Validation run	Shows the first read-only output with source references and reviewer comments.
Run record	Captures tool calls, evidence, assumptions, final decision, and feedback.

Naming

Use FactVerse AI Agent as the product name for agent workflows, customer-facing guidance, and integration documentation.