AWP PLATFORM / ARCHITECTURE

A layered architecture
for a governed workforce.

AWP connects enterprise source systems, domain data, evidence, engineering capabilities, Digital Employees, skills intelligence, orchestration and a unified workspace. Governance, security and compliance apply across every layer.

Enterprise-controlledOSDU-compatible data foundationMCP-enabled capabilitiesHuman oversight
THE AWP FUNCTIONAL STACK

Eight connected layers.
One operating environment.

The architecture diagram supplied for the Oil & Gas Digital Workforce Workspace is represented below as a navigable layered view. The industry-specific data and Digital Employee examples shown here follow that source architecture.

UNIFIED INTERFACEOil & Gas Digital Workforce Workspace
DashboardDigital EmployeesLive InvestigationSkills & CompetencyApprovals & AuditSettings
8User Layer
Unified Digital Employee Workspace

Dashboard · Digital Employees · Live Investigation · Skills & Competency · Approvals & Audit · Settings

7Task Orchestration & Lifecycle Management
Workforce coordination

Task intake · Skills requirement identification · Digital Employee selection · Planning & execution · Evaluation · Human approval

6Skills Intelligence
Skills and competency engine

Skills taxonomy · Skills database · Competency mapping · Proficiency levels · Certification · Training recommendations · Gap analysis

5Industry Digital Employees
Domain workforce packs

Oil & Gas: Seismic · Well Logging · Drilling · Reservoir · Production · Assets

4MCP Gateway & Engineering Capabilities
Controlled capability access

Capability routing · Tool authorisation · Domain tools & calculations · Data retrieval & integration · Result handling & provenance · Permissions

3Knowledge & Evidence Fabric
Evidence-backed intelligence

Evidence extraction · Source references · Provenance · Cross-domain context · Findings & insights · Audit trail

2OSDU-compatible Data Foundation
Domain-aligned information

Well & wellbore data · Seismic/geophysical data · Well logs · Drilling reports · Reservoir data/models · Production time series · Asset/maintenance data

1Source Systems & Infrastructure
Enterprise and industrial sources

Operational documents · Engineering datasets · Enterprise systems · Databases · File repositories · Industrial systems / historians · External sources

Cross-cutting governance: Identity, access, role permissions, data governance, model and tool governance, policy, audit, traceability and responsible AI are not isolated modules. They govern the full lifecycle and every execution path.

TECHNICAL STACK

From user experience
to industrial data.

User layer
AWP web application · Digital Employee Workspace · Monitoring & Insights · Admin Console

Guided workflows, natural-language interface, dashboards, reports, user and policy management.

Application layer
Digital Employee Factory · Blueprint · MCP Provider Selection · Knowledge Binding · Target Process Context

Provider health, tool discovery and test, employee capabilities, live investigation, evidence explorer, investigation trace, evaluation and audit governance.

AI & orchestration
Agent Orchestration Engine · LLM Services · Tool & Capability Manager · Context & Memory

Understands request → plans → selects tools → authorises → executes → collects evidence → reasons → answers.

Integration layer
MCP Connector · Tool Adapters · Knowledge Connectors

Connects MCP providers, handles tool invocation and responses, and binds documents, manuals, SOPs, ontologies and domain knowledge.

Data sources layer
Industrial Systems · Enterprise Systems · File & Document Stores · Data Platforms · External Services

DCS, historian, SCADA, OPC-UA; ERP, CMMS, LIMS, MES, EAM; procedures, manuals, SOPs, data lakes and approved external services.

EXECUTION LOGIC

Understand. Plan. Authorise.
Execute. Evidence. Evaluate.

01User request
02Understanding & target resolution
03Plan generation
04AWP authorisation
05MCP execution
06Evidence capture
07Reasoning
08Final answer

The functional stack specifies that an answer should present findings, list investigation leads, state what is not known and recommend next steps. Tool status and metadata, source references and evidence are retained as part of the investigation record.