FactoryPulse: Operations Overview

Checking Kafka

A live manufacturing control tower powered by a continuous Confluent data backbone.

UPDATED —
Live · LINE-01 / CNC-03
Demo scenario
Factory state

Live production intelligence

LOW
—
Projected output / 1,000 target
CNC-03 Waiting for data
LINE-01 Gear Assembly
Continuous RAG
0
Indexed chunks
Continuous index
Machine condition
— mm/s
CNC-03 vibration
Production
— s
Current cycle time
Baseline 42 s
Quality
—%
Defect rate
Quality risk: —
Exception-to-action

What the Streamhouse sees now

LIVE
MachineHealthy
ProductionOn target
QualityNormal
Recommended action

Continue monitoring.

Recent exceptions

Operational timeline

Continuous RAG

Ask using knowledge that updates as the factory changes.

Maintenance notes, SOPs, and work-order resolutions are streamed into Kafka, chunked and embedded continuously, then retrieved against the latest question.

Knowledge freshness Live 0 chunks currently indexed
Ask FactoryPulse

Grounded maintenance assistant

Evidence-first response
Grounded response

Answer

— ms
Ask a question to see the latest indexed factory knowledge used as evidence.
Retrieval: — Generation: —
▶ Data trace — how this answer was built from Confluent
Ask a question to see the data trace.
Retrieved evidence

What grounded the answer

No evidence retrieved yet.
Knowledge stream

Add knowledge and watch RAG refresh continuously.

Each document becomes an event on rag.knowledge.raw. The indexing path updates without a nightly rebuild.

Publish document

New maintenance knowledge

Publishes to Kafka immediately
Indexed sources

Current knowledge documents

0
Architecture

Continuous RAG as a Streamhouse workload.

The architecture keeps operational state and enterprise knowledge moving on the same governed event backbone while serving both live actions and historical analytics.

Factory sources
PLC / Machine Sensorstelemetry
MES / ERPproduction & orders
Quality Systeminspection results
Maintenance / SOPswork orders & knowledge
→
Confluent Cloud Streamhouse
Kafkadurable event backbone
Flinkjoin · detect · chunk · embed
Schema Registrycontracts & governance
TableflowIceberg / Delta materialization · factory.state & factory.exceptions → historical tables
→
Consumers
Continuous RAGlatest knowledge
Operations UIlive factory state
Alerts / Actionsexception-to-action
Analytics / AIhistorical tables
1Change happensmachine, work order, SOP
2Kafka captures itdurable event
3Knowledge refresheschunk + embedding
4RAG retrieves itfresh evidence
5Factory actsoperator + maintenance
Live events

See the business change as events.

This feed is populated from the same state used by the UI and knowledge pipeline.

Event stream

Most recent events

LIVE
Runtime

Configuration status

Secrets are never returned by the API. This page only shows whether required configuration is present.

Kafka—

Bootstrap server + Kafka API key/secret

Schema Registry—

Auto-activated when SR credentials are present

RAG pipeline—

local demo worker or Confluent Flink

Generation—

Grounded template or watsonx.ai

Tableflow—

Iceberg / Delta materialization for factory.state and factory.exceptions

Useful endpoints

API surface

GET  /api/health
GET  /api/dashboard
GET  /api/tableflow/status
POST /api/knowledge
POST /api/rag/ask
POST /api/demo/seed-knowledge
POST /api/demo/step/{baseline|degrade|critical|resolved}
GET  /docs