Bodegas Bordino uses AI to automate order and delivery-note processing
Beverage distribution for hospitality
Context
Bodegas Bordino started in the early 1990s as a neighbourhood wine shop, also specialising in spirits, that supplied the founder's contacts in the hospitality trade. In 2006 the two brothers who now run it as partners chose to professionalise the business rather than close it: they moved to a warehouse on an industrial estate and have kept growing ever since, with more growth planned. That growth brings more data and more repetitive admin work, and it is what led them to bring AI into their operations.
Challenges
The bottleneck was the time sales reps spent keying in orders and entering them into the system, time they needed to spend with customers. Orders arrived by email, WhatsApp or phone, and the tool that pushed delivery notes and invoices into the ERP relied on rigid templates and often failed: when a document didn't come through, the rep called the office and someone had to email it over. The challenge was to remove that admin load without losing control over each order.
Solution
Adoptian and Bordino built a custom AI platform together, focused on the two most time-consuming flows. For orders, incoming order emails are forwarded to an AI-connected inbox that interprets them and brings them into the platform, where the order team reviews each one, checks for discrepancies and validates it. For documents, delivery notes and invoices, on both the purchasing and sales side, are scanned with a phone, tablet or scanner; the platform matches their lines within minutes and keeps the data available for later queries. The project has moved forward iteratively, through testing and adjustments alongside the Bordino team.
Technical Architecture
The solution combines AI-based data extraction, a bespoke web application and human validation at the critical step: every order is reviewed and validated by the order team before it goes through. AI gives the ERP the flexibility it lacks on its own, compared with the rigid, template-based tool the team relied on until now. The intelligence layer runs on Microsoft Azure AI Foundry, with a multi-agent architecture in which each agent handles one part of the document process.
Results
The order team estimates the platform has made them 70–80% more productive, and sales reps no longer handle the admin side of order entry. A delivery note or invoice is matched in minutes, and documents come through with far fewer errors. “This way, we've cut down on time and errors,” says Jorge Bordino, partner at the company. Next up: querying the documents, collections management and customer reports.
