19 Use Cases. One Predictive Database.
The ERP that learns, built on Aito's _predict, _estimate, _relate, _evaluate and _search
Procurement, intelligence, products, services — every view in this open-source reference is built on Aito's predictive operators. Three industry profiles (industrial maintenance, multi-channel retail, professional services) drive the same code.
ERP Features Grouped by Function
Procurement, intelligence, product/pricing, services. Pick a feature to read its implementation guide on GitHub.
Procurement
Predictive routine that pre-fills your AP team's morning — POs route themselves, smart entry predicts four fields from one supplier pick.
PO Queue
Predicted account, cost center, approver — visualised in three confidence tiers
See in demo →Source →
Smart Entry
Pick a supplier, four fields predict in parallel with cross-highlight
See in demo →Source →
Approval Routing
Predicted approver and approval level via _predict, plus fixed escalation rules; nothing becomes policy without explicit signoff
See in demo →Source →
Automation Overview
Per-field accuracy against baseline, calibration by confidence band, and a learning curve from routed_by × order_month
See in demo →Source →Intelligence
Anomalies, supplier risk, pattern discovery — surfaced, not coded.
Anomaly Detection
Inverse prediction. Three types — amount spike, unknown vendor, mis-coded account
See in demo →Source →
Supplier Intel
Spend leaderboard plus delivery-risk discovery via _relate
See in demo →Source →
Rule Mining
High-lift patterns ranked for review, each with its confidence and evidence
See in demo →Source →
Recommendations
Bought together: _relate lift over order baskets, with the counts behind it (Aurora tenant)
See in demo →Source →Product & Pricing
Catalogue gaps, fair-price bands, seasonal demand, replenishment.
Catalog Intelligence
Multi-field gap-fill on workflow-blocking products — bulk apply across the catalog
See in demo →Source →
Price Intelligence
Fair-price band + Purchase Price Variance with annualised exposure
See in demo →Source →
Demand Forecast
Seasonal blend, same-month aggregation, stockouts-prevented and excess-avoided in €
See in demo →Source →
Inventory Intelligence
Critical / Low / OK / Overstock with tied capital and weekly margin at risk
See in demo →Source →
Invoice Matching
Each invoice line gets five ranked catalogue SKUs from one _predict on the sku link, for a clerk to confirm (Aurora tenant). On the published line-matching benchmark over this demo's synthetic catalogue, Aito ranks the right SKU first 92.4% of the time against 89.9% for a fair BM25 baseline, and BM25 is ahead in the top five, 98.5% against 96.9%.
See in demo →Source →Services & Operations
Project success, utilisation, capacity — beyond procurement.
Project Portfolio
_predict success per active project with $why factor decomposition
See in demo →Source →
Utilization & Capacity
Per-consultant load + role forecast via _predict allocation_pct (Studio tenant)
See in demo →Source →
Project Plan
Phases, tasks, assignees, days and cost drafted with _predict over past tasks; each candidate assignee shows its predicted P(success)
See in demo →Source →
Engagement Planner
Role mix and staffing via _predict on assignments, delivery effort via _estimate, and P(win) plus the likely loss_reason via _predict on quotes
See in demo →Source →
Revenue Outlook
Order book spread by percentage of completion, then shifted by _predict on_time per in-flight project, with on_budget shown beside it
See in demo →Source →Implementation Details
Explore the source, the three tenant profiles, and the architecture decisions
{
"from": "purchases",
"where": {
"supplier": "Lindström Oy",
"description": "Workwear order"
},
"predict": "cost_center",
"select": ["$p", "feature", "$why"]
}
Why This Pattern Works for ERP SaaS
One predictive database, three industry profiles, every override learned from instantly
SQL-Like Predictions
- _predict, _relate, _search, _recommend
- No feature engineering required
- Confidence + $why on every prediction
Three Profiles, One Codebase
- Metsä Machinery (industrial maintenance)
- Aurora Retail (multi-channel commerce)
- Vire Consulting (professional services)
Production-Ready Architecture
- Next.js 16 App Router + FastAPI
- 2-layer cache, schema link traversal
- 26 pytest tests, TypeScript strict
Evaluate in 15 Minutes
From live demo to your own data in three steps
Address
Episto Oy
Putouskuja 6 a 2
01600 Vantaa
Finland
VAT ID FI34337429
