---
title: "Agents for production lines: Trusted decisions in real time"
description: "ProdLine CoPilot addresses in-shift production-line decisions when equipment faults threaten output and data is split across PLCs, SCADA, MES, ERP, and LIMS. The proposed system streams operational technology data into Databricks, joins it with slower business and quality records, and routes questions to domain specialists that calculate recovery, depletion, quality-risk, and maintenance signals. Its orchestrator reads current Unity Catalog state, uses SQL, Genie, AI Search, Model Serving, and Python-based tools, while MILP, stochastic scheduling, Monte Carlo, Bayesian risk, and Pareto methods provide analytical outputs. Recommendations are tested against 1,000 scheduling scenarios and returned with trade-offs in cost, overtime, throughput, quality, and service, but execution remains with line managers, quality leads, and maintenance. The demo drafts work orders, quality deviations, and schedule updates for approval, with traceability storing inputs, assumptions, constraints, approvers, and outcomes."
---

# Agents for production lines: Trusted decisions in real time

[Databricks](https://yomu.fyi/company/databricks) · Mohammad Khelghati · Jul 29, 2026

**Type:** Problem & solution

## Summary

ProdLine CoPilot addresses in-shift production-line decisions when equipment faults threaten output and data is split across PLCs, SCADA, MES, ERP, and LIMS. The proposed system streams operational technology data into Databricks, joins it with slower business and quality records, and routes questions to domain specialists that calculate recovery, depletion, quality-risk, and maintenance signals. Its orchestrator reads current Unity Catalog state, uses SQL, Genie, AI Search, Model Serving, and Python-based tools, while MILP, stochastic scheduling, Monte Carlo, Bayesian risk, and Pareto methods provide analytical outputs. Recommendations are tested against 1,000 scheduling scenarios and returned with trade-offs in cost, overtime, throughput, quality, and service, but execution remains with line managers, quality leads, and maintenance. The demo drafts work orders, quality deviations, and schedule updates for approval, with traceability storing inputs, assumptions, constraints, approvers, and outcomes.

## Context

Production-line data is scattered across PLCs, SCADA, MES, ERP, and LIMS, while line managers must make recovery decisions within minutes. Existing reporting and root-cause-analysis workflows can take until the next morning or 24 hours later, and the people needing answers often do not write SQL.

## Approach / What changed

The system combines near-real-time OT ingestion through Zerobus and Delta with governed MES, ERP, and LIMS data in Unity Catalog. An orchestrator routes natural-language questions to specialist agents that use SQL, Genie, AI Search, Python calculations, anomaly detection, and operations-research solvers. Recommendations remain human-approved and produce draft work orders, quality records, and schedule updates.

## Takeaways

- Zerobus Ingest provides push-based, serverless ingestion through gRPC or REST into Unity Catalog Delta tables, with single-digit-second near-real-time latency and no brokers or partitions.
- Specialists call real analytical methods rather than relying only on an LLM, including OR-Tools SCIP MILP, stochastic scheduling, Monte Carlo forecasts, Bayesian quality-risk models, and Pareto loss analysis.
- Approval gates assign recovery decisions to line managers, hold and release decisions to quality, and work scope and timing to maintenance; proposed system write-backs remain drafts rather than automatic control.

**Tags:** [Architecture](https://yomu.fyi/topic/architecture), [Databricks](https://yomu.fyi/topic/databricks), [Streaming](https://yomu.fyi/topic/streaming)

- Source: [Databricks](https://www.databricks.com/blog/agents-production-lines-trusted-decisions-real-time)
- Source URL: https://www.databricks.com/blog/agents-production-lines-trusted-decisions-real-time
- Ingested by Yomu: 2026-08-30T16:52:50.804Z

[Read original post](https://www.databricks.com/blog/agents-production-lines-trusted-decisions-real-time)