A modular ERP for small and mid-sized businesses

AI assistant

An interpreter between people and the system: it proposes, a person confirms, the system posts

1. 1

   Words, a photo or a file

   An order, a delivery or a stock question — no hunting for the right button.

2. 2

   The AI proposes an action

   A draft of what it would do, not a change.

3. 3

   A person confirms

   You see exactly what will happen before anything is saved.

4. 4

   The system posts it

   Under the same rules and checks as manual entry.

5. 5

   Logged, and can be undone

   Every step is recorded: who did what, and when.

The AI cannot write anything on its own. A confirmed action goes through the same checks as manual entry, lands in the log and can be undone. AI spend is capped by a per-company budget.

In the product today

A working product for formula-based production

Grown from a real manufacturer's request, so the hardest parts were built first: formulas, lots and work with outside processors. Stock and lots, purchasing, formula-based production, customer orders, documents, the AI assistant, an accounting connection and a public API are in place.

Stock and lots are the strongest part: a recall is a single query

• Lots with expiry dates; the lot that expires first is picked first. A quarantined lot is never picked automatically.
• Traceability forward and back: from raw material to finished lot to the customers it shipped to.
• Virtual warehouses: in transit, at the processor, work in progress, quarantine.
• On-hand stock is derived from a movement ledger that can only be added to, so any number can be explained and rebuilt as of any date.

Production and purchasing understand formulas: a raw-material shortage turns into draft purchase orders

• Formulas in percentages or quantities, batch size, yield and scrap, versions, substitutes.
• A production order reserves materials, consumes them by lot and creates a finished lot with its genealogy; under-yield and losses are recorded explicitly.
• Purchase orders: minimum-order and pack-size rules, approval above a threshold, partial receipts, matching of order, receipt and bill, freight and duties spread into cost.
• Shortage planning: demand plus minimum stock minus on-hand minus on-order.

A customer order decides where the goods come from: stock, make, buy or send to a processor

• Flow: quote, order, activation, partial shipments, invoice, return. On activation, reservations, production orders and purchase orders are created in one step.
• Customer credit control: warn, require approval, or block.
• Subcontracting: your own materials go to a processor, the processor's rates are applied, its bill is reconciled, and losses are recorded with a responsible party.
• Margin is visible on every order.

Documents follow your own forms, and accounting is connected rather than replaced

• Customer invoices, quotes, bills of lading and packing slips are generated in Excel in the layout you already use, then stored and attached to the record. PDF output is in progress.
• QuickBooks is the first accounting connection: invoices and orders go out, payments and the balance sheet are read back. Every exchange is visible in a log with a retry button. Connecting live QuickBooks companies follows Intuit's review of the app.
• We don't replace accounting; further accounting systems are connected the same way.

Pricing

Built from a base, modules, an industry pack and services: you pay for what you use

Per month, per company, unlimited users.

• Public prices
• Lots and traceability always in the base
• No charges per user, order or warehouse
• No annual lock-in

The only cap is the monthly AI budget, set per company.