Deployment process

From onboarding to live forecasts in four steps

Lean Path is deployed as a managed service. Your team handles planning; we handle data ingestion, model training, and forecast delivery.

01

Connect

Establish data connections

We work with your IT team to set up the data feeds from your Tier-1 customers and from your internal systems. Most connections use protocols already in place: EDI, SFTP, or direct ERP export.

  • EDI 830 and 862 message parsing configured for your customer formats
  • Historical shipment data imported from your ERP (typically 18 to 24 months)
  • BOM data loaded and validated against your part catalog
  • Inventory position feeds set up for net-requirements calculation
Data sources Step 01
EDI 830
ERP export
BOM data
Inventory
Lean Path ingestion
Connections validated and live

02

Calibrate

Train the demand model

We train a demand translation model on your historical data. The model learns the relationship between assembly pull signals at the Tier-1 level and actual parts demand at your level, accounting for lead times, BOM ratios, and seasonal patterns.

  • Model trained on your supply chain, not on shared or generic training data
  • Signal lag and lead time calibration per part number and customer lane
  • Baseline forecast accuracy reviewed with your planning team before go-live
  • Model behavior explained in terms your planners can interrogate
Training phase Step 02
Historical signals
Actual orders
AI model training
Accuracy review
Adjustments
Model calibrated on your supply chain

03

Forecast

Deliver rolling demand forecasts

Once live, the model generates a rolling demand forecast for each part number in scope. Forecasts are updated each time new pull schedule data arrives from your customers, typically daily or weekly depending on your customer's release cadence.

  • Rolling 8 to 13 week horizon updated with each new schedule release
  • Part-number level forecast with net requirements against current inventory
  • Variance alerts when forecast deviates significantly from the prior period
  • Output delivered to your ERP, shared drive, or planning dashboard
Forecast output Step 03
Pull schedule update
Model inference
Week 1-4
Week 5-8
Week 9+
Forecasts updated with each release cycle

04

Act

Plan with earlier information

With a rolling forecast ahead of the confirmed purchase order, your planning team can make procurement and production decisions earlier. The goal is to reduce emergency orders and overstock by giving planners a reliable signal one to three weeks before the order arrives.

  • Planners review the forecast alongside their existing ERP demand plan
  • High-variance parts flagged for manual review each week
  • Safety stock recommendations adjusted based on forecast confidence intervals
  • Model continues retraining on new data to improve accuracy over time
Planning impact Step 04
Forecast signal
Earlier procurement
Fewer emergencies
Reduced overstock and shortages
Planning cycle runs on earlier signals

Onboarding timeline

Typical deployment milestones

Most Pilot deployments reach production forecasts within eight to twelve weeks of contract signature, depending on data availability.

W1

Data audit

Wk 1-2

W3

Connections live

Wk 3-4

W5

Model trained

Wk 5-8

W9

Production forecasts

Wk 9-12

Ready to talk through deployment?

We walk through your data environment and supply chain structure before recommending a deployment scope. No commitment required.

Request Access View Pricing