Loading...
Procurement›PO Queue›... unrouted POs · 47 received today
Live
_predict
12ms
Avg latency
3
Predict fields
supplier × desc × amount
Features
_predict
POST /api/v1/_predict
{
"from": "purchases",
"where": {
"supplier": "Wärtsilä Components",
"description": "Hydraulic seals #WS-442"
},
"predict": "cost_center"
}
// → cost_center: "Production" (p ≈ 0.94)
{
"from": "purchases",
"where": {
"supplier": "Wärtsilä Components",
"description": "Hydraulic seals #WS-442"
},
"predict": "cost_center"
}
// → cost_center: "Production" (p ≈ 0.94)
How it works
Each incoming PO is scored with aito.._predict. For industrial maintenance, the model examines supplier, description, and amount to predict cost-centre, account code, and approver. Predictions for Wärtsilä Components route to Production / account 4220; T. Virtanen signs the typical case. High-confidence predictions auto-code; low-confidence ones queue for review.
Learn more