Real builds, from operations systems to consumer products

A general contractor ran its entire payment-application process by hand: vendor invoices pulled out of email one at a time, coded manually, and assembled into AIA G702/G703 billings in spreadsheets — with no traceable link between an invoice and the contract line it billed against.

In about six weeks, we shipped the core billing system: invoice intake from email, an accounts payable (AP) review queue, allocations against a locked schedule of values, change orders flowing into contract value, and printable G702/G703 billing packages with their supporting invoices.

The system now also reads each invoice PDF with a vision model and matches it to its vendor and project. Auto-routing happens only when every required field clears an 80% confidence gate; everything else goes to a human review queue. Every business change writes an audit record in the same transaction, keeping the record tied to the work.

After launch, we stayed on as the client's real billing rules surfaced — closed-period revisions, fee rounding, and locked-period snapshots — and shipped them across continued delivery cycles. The billing math and print formats are verified correct by the client. In this project, the AI reads and sorts invoices at about 90% accuracy under human review. The manual email-fetch, sorting, and billing-assembly workflow has been replaced, with review and exception handling retained.

A local Thai retail shop tracked stock and daily sales by hand — no reliable way to see what was running low or what had actually sold that day.

We built StockEasy, a mobile-first inventory and sales tracker that lives entirely inside LINE, the messaging app the shop's staff already use — nothing to install. Staff record sales and stock movements from their phones. The system keeps a full movement ledger and sends low-stock alerts and daily summaries back into LINE. Payments are verified server-side before paid access is granted. Multiple shops and staff roles stay isolated and accountable through scoped access and audit logging.

From first commit to a working, payment-integrated product took under six weeks, with automated tests and a CI pipeline behind every release. As the owner's real operating rules surfaced, the data model evolved through tracked migrations. StockEasy is Valo Studio AI's own product, built in direct partnership with the shop owner who supplied the real operating constraints — and it is ready to launch.