Case study / AI
TimbraBot.
AI-assisted bank reconciliation.
TimbraBot is a bank and tax reconciliation platform developed by WASD Studio. It combines rules, AI and human review to match transactions with invoices and preserve supporting evidence.
- Context
- WASD Studio · Custom development
- My role
- Product · Technical decisions · Delivery
- Status
- In operational use
Inside the product
A proposal with supporting evidence.
Product detail with sample data: a proposed match and its amount breakdown. Interface shown in Spanish.
Enlarge screenshot (new tab) 01
The problem
One payment can cover several invoices; one invoice can be paid in installments. The accounting firm needs to connect bank statements with Mexican tax invoices, explain each match and review exceptions.
02
My contribution
My work at WASD covered product definition, technical decisions and delivery, connecting the accounting workflow with the product experience.
03
Rules first, bounded AI
Rules handle cases that can be repeated and audited. AI steps in where context needs interpretation. A match is applied automatically only with high confidence and hard evidence in the bank description; all other matches go to review. This gives automation a verifiable boundary.
Design decisions
Model proposals are validated on the server before being saved. Supporting evidence remains available for the accountant to approve, correct or reject. The WASD case explains the architecture and its limits.