FactVerse AI Agent Capability Map

The MCP tool reference describes individual tools. This map explains how tools, data, digital twins, and product modules combine into capabilities that customers can evaluate, deploy, and govern.

Capability assembly flow

Capability groups

Capability	Connected components	Typical output	Governance concern
Industrial operations copilot	FactVerse Platform, DFS, Inspector, knowledge stores	Situation summaries, asset answers, procedure guidance, task drafts	Tenant context, source traceability, approval for write actions
Predictive maintenance workflow	DFS signals, asset history, Inspector work orders, AI Agent analysis tools	Maintenance risk explanation, suggested checks, work-order preparation	Model confidence, maintenance authority, record retention
Facility and energy operations	Operational twins, meter data, asset metadata, inspection records	Energy-use calculation support, abnormal trend review, evidence organization	Data freshness, regulatory wording, human engineering review
Simulation and Physical AI workflow	Designer scenes, SimReady assets, simulation services, operational constraints	Layout comparison, process rehearsal, robot or equipment training context	Simulation assumptions, versioned assets, validation boundary
Knowledge and document workflow	Manuals, SOPs, customer documents, and tool references	Retrieval, comparison, translation support, structured answer drafts	Access control, citation quality, confidential document handling
Field execution and feedback	Inspector mobile workflows, alerts, work orders, checklists	Dispatch context, inspection guidance, completion records, feedback loops	Operator approval, audit trail, operational safety
Data governance and integration	DFS pipelines, quality checks, source-system connectors	Data readiness review, schema mapping, anomaly flags, integration status	Data ownership, quality gates, pipeline observability
Agent integration tooling	MCP endpoints, scopes, tool references, customer AI clients	Controlled agent access, runtime discovery, integration test paths	Scope design, key rotation, monitoring, lifecycle management

How to read the map

Start from the business workflow, then identify the product modules and tools required to support it. For example, a predictive maintenance workflow normally needs asset context from FactVerse Platform, operational signals from DFS, maintenance history and work-order context from Inspector, and scoped tools exposed through MCP.

Capability assembly pattern

1. Define the operational question or decision boundary.
2. Attach tenant, asset, and permission context.
3. Retrieve relevant data, knowledge, digital twin state, or simulation assets.
4. Generate analysis or a proposed task with source references.
5. Route write actions through approval and audit.
6. Feed completed work, inspection results, and operator feedback back into the system.

Capability naming

In business and integration discussions, use capability names such as "predictive maintenance workflow" or "industrial operations copilot." Use raw MCP tool names for integration behavior, debugging, and API-level references.

Validation questions

Use these checks before describing a capability as deployable:

Question	Good answer
Which workflow owns the capability?	A named operating, maintenance, engineering, or facility workflow.
Which data proves the answer?	Source systems, datasets, documents, scenes, or work records with timestamps.
Which endpoint exposes it?	Base or module MCP endpoint with a minimum scope set.
Who reviews the output?	A named reviewer with authority for the operating decision.
How does feedback return?	Work-order outcome, inspection note, model review, data correction, or knowledge update.