Industrial Physical AI Platform
Built for Industrial Reliability,
Governance, and Scale
A layered Physical AI architecture — knowledge base, model, agents — designed for the most demanding industrial environments.
Enterprise overview
What it takes to run Physical AI in production
This page covers the deployment, integration, security and certification side of AICraft — the architecture your IT, OT and compliance teams need to sign off. The product capabilities themselves are documented on the platform pages.
Platform capabilitiesSystem Architecture
Reference Architecture
A modular, event-driven pipeline that turns raw IT, OT and ET data into an industrial knowledge base, Physical AI predictions and governed action.
Data Sources
OPC-UA / MQTT
Real-time telemetry
Historians / SCADA
Time-series data
APIs & Webhooks
External integrations
Reference Materials
Manuals & specs
Internal Knowledge
SOPs & SME expertise
ERP Systems
SAP, Oracle, etc.
Ingestion Layer
AICraft Beat
Deployed alongside control systems and plant infrastructure, AICraft Beat consumes data from all operational, document, and business sources and securely moves it into the processing layer.
Edge World Models for APC & MPA
Automatic process control · ultra-low-latency inference
Processing Pipeline
Stream Processing
Real-time event processing, anomaly detection, and signal enrichment
Rule Engine
Deterministic rules and domain-specific logic execution
Knowledge Extraction
Extracting knowledge and meaningful signals from IT, OT and ET sources
Foundational Intelligence
Physical AI models
Training and inference for batch and continuous operations.
Forge Batch World Model
Learns batch dynamics, equipment physics and chemistry to simulate batch progression and optimise operating conditions.
Forge Continuous World Model
Learns continuous process dynamics to model evolving plant states, simulate operating changes and optimise steady-state performance.
Predictive Models
Predicts yield, throughput, quality, batch cycle time and equipment breakdowns to guide operational decisions.
Knowledge Store
Unified Knowledge Store
The central intelligence layer where all processed data, learned patterns, and domain knowledge converge. Procedural graphs encode how plant work is actually carried out — step by step, with preconditions and checks — so PilotX agents execute industrial AI tasks with high accuracy.
Semantic Representations
Unified Namespace
Learnt Knowledge Graph
Procedural Graphs
Lakehouse
User Experience Layer
PilotX agents
Explained Physical AI guidance
Predictive Alerts
Instant operator actions
Reports
Automated insights
Proactive Intelligence
Anticipate risks and opportunities
Digital Twins & Simulation
Live equipment and process context
Simulation
What-if runs on the twin
End-to-End Platform Configuration
Configure ingestion, models, workflows and experiences
We own the stack end-to-end
From the models to the user experience layer
Our core IP is in the Forge world models and the PilotX agent models that support every layer above. Because nothing depends on third-party AI services, the full platform runs on-prem — including fully air-gapped plants with no internet connectivity whatsoever.
Data in scope
What the knowledge base holds — and where it stays
Every source below is ingested under your residency boundary, with lineage, role-based access and a full audit trail. Nothing leaves the deployment zone you define.
From these sources the knowledge base builds semantic representations, a unified namespace, a learnt knowledge graph, a lakehouse — and procedural graphs that capture how plant work is done step by step, so PilotX agents carry out industrial AI tasks with high accuracy.
Deployment Options
Your Infrastructure, Your Rules
SaaS
AICraft-managed cloud infrastructure with secure API bridge to plant systems.
Connectivity
Integrations
Connect to any OT/IT data source with pre-built connectors and open APIs.
OPC-UA
Industrial automation protocol
MQTT
Lightweight messaging
HTTPS APIs
REST & webhook integration
SQL Databases
PostgreSQL, MySQL, MSSQL
Historian Systems
OSIsoft PI, Wonderware
SCADA/DCS
Control system connectors
SAP/ERP
Business system integration
Cloud Storage
S3, Azure Blob, GCS
Trust & Compliance
Security & Governance
RBAC & Separation of Duties
Granular role-based permissions with least-privilege access. Separate admin, operator, and viewer roles.
Encryption
TLS 1.3 for data in transit. AES-256 for data at rest. Key management with customer-controlled options.
Audit Logging
Immutable audit trails for every user action, data access, and system event. Exportable for compliance.
Network Segmentation
OT/IT/DMZ isolation patterns. Support for Purdue model and IEC 62443 zone architectures.
Data Governance
Data classification, retention policies, and lineage tracking. GDPR and industry-specific compliance.
Compliance Certifications
ISO 27001, ISO 9001, and ISO 42001 certified. Regular third-party security audits.
Certified & Compliant
ISO 27001
Information Security Management
ISO 9001
Quality Management System
ISO 42001
AI Management Systems