HFS
Hybrid Flow System
Compose triggers, data processing, model recommendations, human approval and governed actions into an inspectable operational workflow.
The problem
Automation often remains scattered across scripts, schedules and personal judgment. When AI is added without state, permissions and evidence, operational risk grows.
The outcome
Preserve every input, decision, approval, execution and outcome in a node-based workflow so the capability can be repeated and improved.
Operational context
Make data, models, approvals and controls one inspectable flow.
HFS keeps triggers, evidence, rules, AI options, human approval, controlled execution and outcome learning in one workflow.
- 01Visual workflowsExpress triggers, conditions, data, algorithms, human approval and outputs through nodes and connections.
- 02Built-in anomaly detectionEmbed statistical or algorithmic nodes while preserving provenance and downstream handling boundaries.
- 03Isolated C# scriptingAllow constrained extension while governing execution time, resources, dependencies and callable scope.
Core capabilities
From product module to reusable operational capability.
The codebase includes a workflow engine, anomaly detectors and a script host. The visual designer, node catalog and production isolation must be verified against the released version.
Visual workflows
Express triggers, conditions, data, algorithms, human approval and outputs through nodes and connections.
Built-in anomaly detection
Embed statistical or algorithmic nodes while preserving provenance and downstream handling boundaries.
Isolated C# scripting
Allow constrained extension while governing execution time, resources, dependencies and callable scope.
Governed actions
Add allowlists, simulation, dual approval, timeout, recovery and complete records to high-risk actions.
Data and responsibility flow
Every step stays visible, verifiable and governable.
- 01Start from an event, schedule or person
- 02Collect real-time and historical evidence
- 03Use rules, algorithms or AI to form options
- 04Execute after human approval and collect the outcome
Use cases
Start with one high-value loop.
- Anomaly classification and assignment
- Energy strategy recommendation and approval
- Quality-event handling
- Operational orchestration across OT and IT
Responsibility boundary
HFS must not bypass existing control responsibility. Every external write and AI recommendation must distinguish recommendation, approval, dispatch, execution and outcome.
Current technical baseline: The codebase includes a workflow engine, anomaly detectors and a script host. The visual designer, node catalog and production isolation must be verified against the released version.