PO intelligence module
Procurement intelligence beside the ERP, not inside it. Consolidates duplicate demand across plants, scores supplier choice on actual history, benchmarks price and lead time, detects match exceptions - and proposes, never writes.
flowchart TB
subgraph Typical["Typical enterprise approach"]
direction LR
T1[ERP core]:::erp --> T2[Custom fields, workflows, scripts]:::risk --> T3[Upgrade risk, vendor lock-in]:::risk
end
subgraph RCT["RCT approach"]
direction LR
R1[ERP core]:::erp --> R2[Read-only extract]:::module --> R3[PO intelligence layer]:::module --> R4[Proposal queue]:::outcome --> R5["Buyer approval -> optional write-back"]:::outcome
end
T1 -. same unmodified ERP .- R1
classDef erp fill:#fff4df,stroke:#b36b12,color:#3c2a10;
classDef risk fill:#fff0f0,stroke:#b84a4a,color:#581d1d;
classDef module fill:#eef2ff,stroke:#4c63c7,color:#1e2d75;
classDef outcome fill:#f2f7e8,stroke:#6d8f26,color:#26380d;
Same ERP, zero schema or workflow changes underneath. The intelligence layer reads, proposes, and stops - the buyer decides.
- Integrated pipeline
- Plant and Procurement Intelligence Pipeline
- Role in stack
- ERP-side intelligence -> proposal only
- Current state
- Concept mock
- Best demo use
- Opens the procurement-beyond-ERP conversation and the CIO-safe read-only deployment argument
- Honest boundary
- No code exists yet; built 2026-09-13 to scope fitment
Seven stages of one PO review cycle: read-only ERP extract, cross-plant demand consolidation, supplier selection scoring, price and lead-time benchmarking, three-way match exception detection with routing, a buyer approval gate, and a proposed write-back with an audit record.
Completing the run opens the order review queue — six orders filterable by exception type, each expanding into the finding, the evidence that produced it, and the recommended action.
The demo is a static mock. It runs entirely in your browser on synthetic data — nothing is gated, nothing is collected.
See the shared reasoning core — sim mode ↗ — the same classify/validate/traverse/evidence pattern behind this product's narrative layer, running live on a small invented dataset.