Industrial Diagnostics
Industrial Edge Discovery
Diagnostics and Live System Documentation
A read-only platform for industrial edge support teams. It connects to remote devices, discovers running services and dependencies, generates live documentation, and shortens root-cause analysis with snapshot-based diagnostics.
4
Architecture Layers
2
Operating Modes
6
Diagnostic Steps
Preview
Device Snapshot
edge-berlin-03
Ubuntu 22.04 · uptime 68d · VPN + SSH
opcua-collector
Healthy · last ingest 34s ago
mqtt-bridge
Warning · broker handshake failed
timescaledb
Healthy · last write 3m ago
Likely fault domain
Upstream broker connectivity after recent endpoint change
Snapshot diff found a new MQTT endpoint and no collector-side regressions.
01
Problem
Industrial edge deployments drift over time. Containers get added, configs change, documentation falls behind, and support engineers often begin incidents without a reliable picture of what is actually running on the target device. Troubleshooting becomes slow, expert-dependent, and inconsistent across customers.
Support Starts Blind
Incidents begin without a trustworthy view of services, ports, logs, or recent changes on the target device.
Documentation Drifts
Real deployments evolve after go-live, but drawings and handover docs rarely keep pace with brownfield changes.
Expert Bottlenecks
Troubleshooting depends on whichever engineer remembers the customer-specific setup from six months ago.
02
System Flow
The core value is not a chatbot. The platform works by securely collecting technical facts, building a device model, and then using that model to generate documentation and guided diagnostics.
Access Layer
Remote Edge Device
Approved VPN, SSH, or customer remote-access tooling establishes read-only access.
Docker containers
System services
Logs and configs
Databases and ports
Knowledge Model
Inspect, classify, understand
Discovery Engine
Normalizes raw inspection data from the device
Role Classifier
Identifies collectors, dashboards, databases, and unknowns
Snapshot Diff
Compares current state against previous and known-good baselines
Live Dossier
Auto-generated documentation with service inventory, topology, and storage paths.
Health Summary
Detects stale collectors, stopped services, and disconnected brokers.
Guided Diagnostics
Highlights likely fault domains and the next logs or configs to inspect.
Step 1
Connect securely
Step 2
Inspect system state
Step 3
Classify components
Step 4
Generate dossier
Step 5
Compare snapshots
Step 6
Guide diagnostics
Brownfield Mode
Connects into messy, already-running customer systems and infers structure from what is actually deployed.
Useful immediately on existing fleets
Handles incomplete standards and unknown services
Ideal for audits, support, and handovers
Greenfield Mode
Uses naming conventions, labels, and health endpoints from standardized deployments to improve precision.
Higher-confidence classification
Cleaner documentation and automation
Better long-term change tracking
03
Technical Approach
Designed the platform around deterministic discovery rather than a chat-first UX. A secure access layer connects over approved remote methods, the discovery engine inventories Docker, services, logs, configs, databases, and network state, and a knowledge model turns raw inspection into a structured system view. Versioned snapshots enable drift detection and guided diagnostics, while AI remains an optional explanation layer on top of a rules-driven foundation.
{
"device": "edge-berlin-03",
"services": [
{ "name": "opcua-collector", "role": "collector", "status": "healthy" },
{ "name": "timescaledb", "role": "database", "status": "healthy" },
{ "name": "mqtt-bridge", "role": "bridge", "status": "warning" }
],
"drift": [
"mqtt endpoint changed",
"new collector-debug container"
],
"diagnostics": {
"faultDomain": "upstream connectivity",
"nextChecks": ["mqtt-bridge logs", "broker endpoint config"]
}
}Design Principle
Reliable inspection comes from evidence first. AI is only useful after the platform has mapped the real system state.
Deterministic discovery first, AI explanation second.
Read-only by default
No automatic config changes
Auditable checks and evidence trails
Customer-controlled credentials
04
Platform Modules
Module 1
Connector Module
Module 2
Discovery Module
Module 3
Parser and Classifier
Module 4
Knowledge Model
Module 5
Snapshot and Diff
Module 6
Diagnostics Engine
Module 7
Report Generator
Module 8
Optional AI Assistant
05
MVP Scope and Outcomes
Recommended MVP
Connect to one industrial edge device and generate a useful system report plus basic diagnostics.
SSH-based access to the target device
Docker, service, config, log, and database discovery
Basic role classification and health summary
Versioned snapshots and snapshot comparison
Generated technical dossier for support handover
Outcome 1
Defined a four-layer architecture from secure access through diagnostics
Outcome 2
Scoped a read-only MVP around discovery, generated dossiers, and snapshot comparison
Outcome 3
Modeled brownfield and greenfield operation for messy and standardized fleets
Outcome 4
Separated deterministic discovery from optional AI explanations for industrial trust
Outcome 5
Designed drift detection against previous and known-good snapshots
06