Flower & plant logistics · Netherlands
The whole export paperwork chain for flowers and plants, in one application
The situation
A Dutch logistics company trucks flowers and plants from growers to customers outside the EU. Each truck carries goods from many suppliers, and each shipment needs an export declaration, transit documents and proof of exit. All of it ran on email and spreadsheets.
What we built
A web application, built from scratch, that runs the whole chain: AI reads supplier invoices, loads are grouped into truck runs, export declarations go to the customs broker, and every MRN is tracked in the EU system until exit is confirmed.
What changed
Operators work from one screen per truck instead of a pile of emails, and the company sees at a glance which shipments are still waiting for confirmation of exit.
- Mail intake
- AI invoice reading
- Declarations and manifests per run
- MRN tracking and alerts
- Portals for the broker, suppliers and client
- Run accounting
Wholesale distribution · Latvia
AI inside the client's existing ERP
The situation
A Latvian wholesale distributor had run the same ERP for more than ten years. Customer orders arrived by email and as PDFs and were keyed in by hand, and the sales team spent a good part of each day looking up stock, prices and delivery dates for customers.
What we built
An AI layer connected directly to the ERP. Incoming orders are read and turned into draft sales orders for a person to confirm, and sales staff can ask plain-language questions about stock, prices and order status. It all runs on the company's own servers, and the ERP itself wasn't changed.
What changed
Orders are no longer retyped by hand, and the sales team gets answers without digging through ERP screens. The ERP vendor and the existing processes stayed the same.
Accounting firm · Rīga
Client paperwork in, tax declarations out
The situation
A Riga accounting firm received its clients' invoices, receipts and contracts by email, in shared folders and on paper. Filing declarations in the State Revenue Service's EDS and checking official letters in e-adrese took hours of copying between systems every month.
What we built
An intake layer that sorts incoming documents by client and type, pulls out the figures the accountants need and flags anything unusual for review. Declarations are prepared and submitted to EDS straight from the firm's own system, and e-adrese notices are matched to the right client.
What changed
Accountants start the month with their documents already sorted, declarations go out from the system they work in, and nothing from e-adrese gets missed.
Retail · Regional chain
Helpdesk triage that never leaves the client's servers
The situation
A retail chain's internal helpdesk, serving stores, the warehouse and head office, handled a steady stream of tickets, many of them repeats, all sorted by hand.
What we built
A triage layer running on the chain's own servers that categorises each ticket, routes it to the right team and suggests answers from past resolutions. No ticket data goes to an outside AI provider.
What changed
Tickets reach the right person sooner, repeat questions get answered from what the team already knows, and customer data stays inside the company.
Your turn
Got something similar?
Paperwork nobody wants to type, questions your team answers every day, a system you can't simply replace. That's usually where we start.
Client references are available during procurement, subject to confidentiality.