How Conduit Works
A federated architecture that queries data where it lives. No data lakes, no ETL pipelines—just intelligent query routing.
Architecture
Federated by Design
Control Plane orchestrates context and queries. Edge Connectors execute at the source.
From Question to Answer
Four steps from natural language query to unified results.
Connect Your Sources
Deploy edge connectors near your data sources. Each connector runs as a lightweight container or native binary, connecting to your historians, SCADA systems, and PLCs.
- Support for Ignition, PI Historian, OPC-UA, Modbus, MQTT
- Connectors register automatically via MQTT
- Local SQLite cache for offline resilience
- Minimal footprint—runs on edge hardware
Discover & Contextualize
Conduit discovers tags from your connected sources and builds a unified catalog. Add descriptions, owners, relationships, and semantic context.
- Automatic tag discovery from all sources
- Vector embeddings for semantic search
- Dynamic UNS built from usage patterns
- ISA-95 hierarchy mapping
Query in Natural Language
Ask questions naturally. AI interprets your intent and generates a structured query you can review before execution. Queries are dispatched to edge connectors in parallel.
- Voice-first, hands-free operation
- AI-generated queries you review and approve
- Cross-source correlation in single queries
- Full audit trail of every query
Get Unified Results
Edge connectors translate queries to native formats, execute at the source, and return results. Data never moves—only queries and responses traverse the network.
- Parallel execution across sources
- Partial results on source failures
- Sub-second response times
- Results cached in pattern cache for trends
Query Compilation Flow
From natural language to native database queries.
Ready to see it in action?
Get a personalized demo of Conduit's federated architecture with your data sources.