Allodial PredictAllodial Predict

Methodology

From messy historical data to reorder intelligence.

Six steps from raw ERP export to a live operational risk dashboard, with a replay validation before you commit to anything.

01

Import your distributor data

Connect QuickBooks, Sage, or your existing ERP, or drop a CSV. Allodial ingests customers, products, inventory counts, and purchase/order history. No reformatting required.

Supported: ERP API, CSV/SFTP, QuickBooks export, manual entry
02

Clean and map the data

A guided wizard validates your import, suggests field mappings, flags missing or inconsistent values, and lets you correct them before analysis begins.

Validation: duplicate SKU detection, date parsing, unit normalization
03

Run a What-If Replay

Before going live, simulate the model against your historical data. See what it would have caught, what it would have missed, and the estimated dollar value of the difference.

Output: catch rate, precision, F1 score, estimated upside
04

Tune thresholds before going live

Review model recommendations product-by-product. Lead time is the only manual input. Everything else is calibrated from your usage history.

Only operator input required: delivery lead time per product
05

Monitor live reorder risk

The dashboard surfaces severity-tagged alerts (critical → low) for every account and SKU. Reorder windows update automatically as new counts are logged.

Alert severities: Critical / High / Medium / Low
06

AI-assisted summaries

Replay narratives and account summaries are written in plain English, explaining what the model found and why it matters for your specific operation.

Allobot uses AI to summarize results, sending operational context to OpenAI under its API terms. Your statistical predictions use no external AI, and your data is never used to train AI models.

Data Requirements

What data do I need?

Required
  • Customer accounts
  • Product catalog with units
  • Inventory count history (date + on-hand qty)
  • Order / purchase history
Recommended
  • Lead times per product
  • Location/branch data
  • Historical reorder quantities
  • Delivery confirmation dates
Optional
  • Freight cost per order
  • Supplier lead time variability
  • Seasonal demand flags
  • Customer tier / priority

Get Started

Ready to run a Replay on your data?

We'll run the Replay and walk you through what the model finds.