mcp · model context protocol
tools are not the intelligence.the protocol that gives them meaning is.
the studio 2.5 mcp server is the wiring layer that connects every agent, every tool, and every data source to a single coherent reasoning surface. it is not a wrapper. it is an architecture.
the gap
most ai integrations are one offs. a connector to one system. a query against one dataset. an answer with no memory of where it came from. the context evaporates. every question starts from zero.
chapter · 01
one protocol. every source.
the model context protocol is the standard by which agents request and receive context without bespoke glue per integration. pull structured geometry, live operational data, program schedules, sensor streams. through a single call surface. add a new source without rewriting the advisor. the list is not fixed. the protocol is composable.
the full asset intelligence ecosystem. every source your estate already generates.
capability abstraction · solid nodes are platform connectors
world model · geometry & asset model · 8 sources
- kahua
- ifc
- revit
- navisworks
- openBIM / IDS
- bentley openroads / openbridge
- cesium
- point cloud / laser scan
construction & program management · 13 sources
- kahua
- oracle primavera cloud
- oracle primavera p6
- oracle unifier
- ineight
- procore
- autodesk forma
- autodesk construction cloud
- e-builder
- contract management
- risk register
- health & safety
- carbon & embodied carbon
asset & facilities management · 7 sources
- kahua
- IBM maximo
- oracle unifier
- SAP PM / SAP S4HANA
- archibus / eptura
- tririga
- coupa
operational technology & building intelligence · 9 sources
- scada
- building automation system (BAS)
- building management system (BMS)
- energy management system (EMS)
- CMMS / predictive maintenance
- iot sensors & streams
- digital twin platforms
- CCTV & physical security
- air quality & environmental
sustainability & carbon · 5 sources
- carbon accounting
- embodied carbon (LCA)
- energy benchmarking
- ESG reporting
- green certification data
inspection, compliance & quality · 4 sources
- site inspection platforms
- quality management
- compliance & permitting
- as-built & handover records
document & knowledge management · 7 sources
- oracle aconex
- bentley infrastructure cloud
- autodesk forma
- procore docs
- microsoft sharepoint
- opentext / documentum
- project information management
enterprise & collaboration · 7 sources
- microsoft sharepoint
- microsoft teams
- microsoft 365 (m365)
- microsoft fabric
- power bi
- azure data lake / synapse
- SAP S4HANA
spatial, gis & geospatial intelligence · 7 sources
- esri / arcgis
- cesium
- qgis
- mapbox
- digital elevation models (DEM)
- utility network data
- smart city / urban data
chapter · 02
a tool ontology built for infrastructure.
tools in the studio 2.5 mcp server are not generic. each one is authored for the specific language of infrastructure. assets, programs, elements, spaces, costs, schedules, risks, and conditions. the ontology is the vocabulary that makes reasoning about a real portfolio possible.
asset.query
retrieve element level properties, conditions, and metadata
geometry.resolve
resolve element topology, spatial adjacency, and semantic geometry into structured context an agent can reason across
program.schedule
pull live schedule data, milestones, and critical path signals
cost.index
access cost plans, actuals, variances, and forecasts
risk.register
surface open risks, likelihoods, and mitigation status
sensor.stream
ingest real time iot and monitoring data to grounded context
spatial.context
surface geometry grounded context: enclosures, proximities, zone memberships, and change deltas since last sync
document.trace
retrieve referenced documents, specs, and decision records
portfolio.roll
aggregate signals across assets, programs, and geographies
governance.check
validate actions and responses against policy and access controls
audit.trail
return lineage, sources, and assumptions for any answer given
chapter · 03
geometry is not a viewer. it is a reasoning substrate.
the world model 3d as code integration is the geometry processing layer beneath the mcp server. spatial relationships, element topology, bounding conditions, and proximity. all of it resolved as structured data that an agent can reason across. the model does not render for humans. it computes for the advisor.
when an agent calls geometry.resolve, it receives structured spatial context: adjacencies, enclosures, load paths, zone memberships, and change deltas since last sync. this is what makes spatial questions answerable at scale, not just viewable.
chapter · 04
every call is governed. every answer is auditable.
the mcp server does not operate without a governance layer. every tool call is validated against a policy engine before it executes. every response is logged with lineage. access is scoped to roles, assets, and programs. the rest api surface exposes this infrastructure to any authorised system. with full audit trail and schema enforced contracts.
this is not bolt on compliance. the governance architecture is foundational. it is what makes the advisor safe to put in front of decision makers.
govern
policy first execution
every mcp tool call clears a role, scope, and data access policy before it runs. no call escapes the governance surface.
secure
zero trust by default
authentication, authorisation, and audit are not optional layers. they are the default operating posture of the server.
expose
rest api with schema contracts
every capability in the mcp server is also available as a governed rest endpoint with versioned schemas, rate controls, and access scoping.
chapter · 05
the stack from signal to answer.
the mcp server sits at layer five of a seven layer intelligence architecture. below it: physical reality, structured geometry, integration, and the living model. above it: agentic reasoning and the human surface. it is the wiring that makes the layers coherent.
- 01
physical reality
- 02
structured geometry
- 03
integration
- 04
the living model
- 05
MCP orchestration
- 06
agentic reasoning
- 07
human surface
the protocol is live.
the studio 2.5 mcp server is in active deployment. if you are building infrastructure intelligence at scale, this is the layer you need under it.