Technical Overview

The ShopFloor Production Analysis Agent is a conversational AI tool hosted on Aptean's shared agent platform. It provides real-time and historical operational insights via natural language queries. By connecting directly to read-only ShopFloor APIs via the Model Context Protocol (MCP), the agent correlates operator performance, work order progress, offstandard downtime, SMV metrics, and training curves into actionable advice.

Note: The agent is purely analytical and advisory. It provides data-backed suggestions but never modifies system data or updates settings automatically.

Key Scenarios & Capabilities

1. Efficiency Diagnosis

Purpose: Identify root causes when a module, line, or team is underperforming relative to target efficiency.

Sample Prompt: "Why is Module 3 running at 58% efficiency today?"

Information Returned:

Target vs. actual efficiency per module and operator.

Operations furthest from standard time.

Breakdown of operators currently on training curves versus fully trained operators.

2. SMV Accuracy Review

Purpose: Uncover operations where standard minute values (SMVs) consistently deviate from actual cycle times.

Sample Prompt: "Which operations have actual cycle times more than 10% above their SMV across the last month?"

Information Returned:

Operations with ≥5% deviation across 3+ work orders (default window: 30 days).

Average over-run percentage and citing work orders as evidence.

Recommendations on which operations warrant a formal time study.

3. Offstandard Downtime Analysis

Purpose: Rank top downtime drivers and isolate high-offstandard modules or teams.

Sample Prompt: "What caused the most production downtime last week?"

Information Returned:

Categorized downtime (machine breakdown, waiting for work, training, absence).

Total minutes lost ranked by category.

Breakdown of self-reported vs. supervisor-approved offstandard time.

4. Training Curve Identification

Purpose: Distinguish expected learning-curve underperformance from skill/operational issues.

Sample Prompt: "Are any underperforming operators on Line 2 still on their training curve?"

Information Returned:

Active training curve progress (e.g., Week 3 of 8) and target pace.

List of underperforming operators who have completed their training curve.

5. Work Order Progress Tracking

Purpose: Project work order completion times based on real-time operator pace.

Sample Prompt: "Will we hit today's production target on work order WO-2045?"

Information Returned:

Produced vs. target quantity and remaining units.

Projected completion time based on active operator efficiency.

Warning flags for orders >15% behind schedule.

6. Plant-Level Performance Summary

Purpose: Receive an executive overview of overall shift status.

Sample Prompt: "Give me a summary of today's plant performance."

Information Returned:

Active operator count and overall plant efficiency vs. goal.

Active work order counts and offstandard approval queues.

Top 3 efficiency concerns by module.

Response Guardrails & Data Rules

Evidence-Based Citing: Every response cites specific source entities (Operator IDs/Names, Work Order IDs, Operation Names, and Location/Module IDs).

Scope Security: Users can only query data within their authorized scope using existing bearer-token authentication permissions.

Real-Time vs. Historical Identification: Responses explicitly differentiate between active shift data and historical logs (up to 90 days).

Advisory Limits: For SMV review flags, the agent recommends conducting a time study and provides observed average actual cycle times; it never proposes specific numeric replacement values.

Created with the Personal Edition of HelpNDoc: Make Your PDFs More Secure with Encryption and Password Protection