REGISTER REGISTER TO LEARN

Pull Actuals -> Score Adherence -> Explain Slips -> Flag Downstream Risk

Schedule Adherence and Slippage Explainer

This agent reads your published schedule and the MES batch actuals, then computes how closely the day ran to plan. It names each slipped order, gives the likely reason from the run and cleaning logs, and lists the downstream orders now at risk so planners see the knock-on before the next shift starts.

The Problem

The board said the batch finished at noon. It finished at four. Nobody could say why.

The morning schedule looked fine. By afternoon three orders had slid and the planner is stuck rebuilding the sequence from memory and chat messages. Was it late charging, a long reaction hold, or a cleaning changeover that ran over? By the time anyone reconstructs it, the same slip has already pushed two orders on the next shift.

What Changes

The board says the batch slipped. The agent says why, and what slips next.

The agent lines up planned start and stop against MES actuals for every batch, scores adherence, and tags each slip with the likely cause pulled from charging times, reaction duration, and cleaning logs. It then walks the sequence forward and flags the downstream orders that will miss their window if nothing changes.

Humans stay in control: the agent scores and explains, the planner decides what to reschedule and signs off on the new sequence.

HOW THE AGENT WORKS

TRIGGERLLMAPIDECISIONACTIONWAIT
TRIGGEREnd of shift schedule pullFires when the shift closes and pulls the day's plan and actuals for scoring.
APIFetch schedule and actualsReads the published schedule, MES batch start and stop times, and cleaning logs.
LLMMatch orders and score adherenceLines up each order's planned versus actual times and computes an adherence score.
LLMExplain each slipped orderTags every late order with the likely cause from charging, reaction, and cleaning data.
DECISIONDownstream orders at risk?
LLMRank downstream riskWalks the sequence forward and ranks orders that now miss their window.
ACTIONSend adherence reportPosts the scored, explained report to the planner ahead of the handover meeting.
The agent does not re-plan production or move orders. Your production planner does. The agent removes the after-shift reconstruction so the planner starts from a scored, explained list instead of a blank sequence.

HOW TO BUILD IT

Three levels of depth. Pick the one that matches where you are.

For the ops person who needs to understand what's involved and brief their team.

Schedule and MESPlan export plus batch actuals
The AgentWorkflow engine + LLM
Planner inboxScored report with reasons

WHAT YOUR TEAM PROVIDES ONCE

  • Adherence score - how closely the shift ran to the published plan
  • Slipped order list - each late order ranked by size of miss
  • Likely reason - late charging, extended reaction, or cleaning overrun tagged per slip
  • Downstream risk list - orders on the next shift that now miss their window
  • Plain-word summary - two lines a supervisor can read on the floor
What happens day-to-day: at shift end the agent pulls the plan and the MES actuals, scores adherence, tags why each order slipped, and drops a ranked report in the planner's inbox before the handover meeting.

For the IT person or technical lead who needs to scope the integration.

INQUIRY SOURCES

APS/ERP schedule export (published plan with target start and stop per order)
MES batch actuals (actual charge, reaction, and completion timestamps)
Cleaning and changeover logs (equipment turnaround times between batches)

WORKFLOW ENGINE (N8N)

Shift-end timer fires
Pull schedule export
Pull MES actuals and logs
Match orders and compute adherence
Classify slip reason per order
Any downstream order at risk
Publish adherence report

CONNECTS TO

APS/ERP schedule (SAP / Oracle / AspenTech)Read: planned order start, stop, sequence. Write: none
MES (AspenTech / Rockwell / Siemens Opcenter)Read: batch actual start, stop, phase times. Write: none
CMMS / cleaning log (SAP PM / IBM Maximo)Read: changeover and clean-in-place durations. Write: none
Notification (Slack / Teams / Email)Read: none. Write: adherence report, slip list
WITHOUT PMS API

Without API access, export the schedule and MES actuals to CSV and drop them in a watched folder. The agent scores and explains from the files, but someone pulls the exports each shift and cleaning times may be entered by hand.

WITH PMS API

With API access the agent reads the published plan and MES actuals directly the moment the shift closes, so the report is ready before the handover. Timestamps come straight from the historian with no retyping. Best experience.

For the person who wants to build this agent step by step. Each node below maps to an n8n workflow node.

TRIGGERNode 1
Trigger - fires at shift end
Schedule trigger set to each shift boundary, or a webhook from the MES when the last batch of the shift reports complete.
Output: A run event carrying the shift window start and end times.
APINode 2
API - fetch schedule and MES actuals
HTTP request nodes to the APS/ERP schedule export and the MES history for the shift window, plus a read of the cleaning log.
Output: Planned orders with target times and actual batch timestamps for the window.
LLMNode 3
LLM - match orders and score adherence
Model node with a low temperature and the planned and actual arrays passed in as JSON.
Output: An adherence score and a per-order slip in minutes.
You are a production schedule analyst. Compare the planned schedule to the MES actuals below. For each order compute the slip in minutes (actual stop minus planned stop) and an overall adherence score from 0 to 100 where 100 means every order finished on plan. Planned: {planned_orders} Actuals: {actual_orders} Respond ONLY with valid JSON. No explanation.
INPUT
{"planned_orders": [{"order": "B1042", "planned_stop": "12:00"}], "actual_orders": [{"order": "B1042", "actual_stop": "16:00"}]}
OUTPUT
{"adherence_score": 74, "orders": [{"order": "B1042", "slip_minutes": 240, "status": "late"}]}
LLMNode 4
LLM - explain slips and rank downstream risk
Model node fed the slipped orders, phase timings, and the remaining sequence.
Output: A reason per slip and the downstream orders now at risk.
You are a production planning assistant. For each slipped order below, pick the most likely reason from late_charging, extended_reaction, cleaning_overrun using the phase and cleaning data. Then list any downstream orders whose planned start falls inside the delay window. Slipped: {slipped_orders} Phases: {phase_data} Downstream: {downstream_orders} Respond ONLY with valid JSON. No explanation.
INPUT
{"slipped_orders": [{"order": "B1042", "slip_minutes": 240}], "phase_data": [{"order": "B1042", "charge_delay_min": 30, "reaction_over_min": 180, "clean_over_min": 30}], "downstream_orders": [{"order": "B1051", "planned_start": "13:00"}]}
OUTPUT
{"slips": [{"order": "B1042", "reason": "extended_reaction", "at_risk_downstream": ["B1051"]}]}
ACTIONNode 5
Action - post adherence report
Notification node that formats the score, slip list, and risk list into a message to the planner channel.
Output: A delivered report ready for the handover meeting.
MVP vs Advanced — Where To Start
MVP — START HERE
  • Start with one production line and one shift so the plan and actuals stay small.
  • Pull the schedule export and MES actuals as CSV into a watched folder.
  • Match orders on order number and compute the slip in minutes.
  • Score adherence as the share of orders that finished within their window.
  • Tag slip reason using three buckets: late charging, extended reaction, cleaning overrun.
  • Email the ranked report to the planner and read it out at the next handover.
ADVANCED — ADD LATER
  • =+ Connect the MES historian API so actuals load the second the shift closes.
  • =+ Add clean-in-place durations from the CMMS to separate cleaning overruns from reaction holds.
  • =+ Walk the full sequence forward to flag every downstream order at risk, not just the next one.
  • =+ Track adherence per line and per product over weeks to surface repeat offenders.
  • =+ Post the report to a Teams channel with the top three slips pinned.
  • =+ Feed confirmed slip reasons back to tune the classifier for your process.

Full Video Walkthrough Coming Soon

Watch the agent turn a shift of MES timestamps into a ranked, explained slippage report.

Register to be notified

Everything on this page is free to learn.

When you're ready to build — do it yourself or bring us in.

Register FOR WALKTHROUGHS
Book a call to scope this

OTHER AGENTS IN LEASING & MARKETING

Frequently asked questions

Schedule Adherence and Slippage Explainer is an AI agent that compares planned schedule to actual completions, scores adherence, explains every slip.
The board said the batch finished at noon. It finished at four. Nobody could say why. The Schedule Adherence and Slippage Explainer addresses this: compares planned schedule to actual completions, scores adherence, explains every slip.
No. Your team stays in control. The agent does not re-plan production or move orders. Your production planner does. The agent removes the after-shift reconstruction so the planner starts from a scored, explained list instead of a blank sequence.

This is 1 of 1 agents in Production Planning & Control. → · Explore all 15 departments →