Loading experience
Three mechanics and one hundred trucks. A Uruguayan industrial group took fault diagnostics from days to minutes without replacing a single person.
Trucks diagnosed in minutes by three mechanics
A Uruguayan industrial group with 100 trucks across three countries took fault diagnostics from days to minutes without replacing a single person. One operator, inside the team, one process at a time.
Fault-diagnostics turnaround
Outside repairs per month
Admin workload carried by AI
Belmar grew faster than it formalized. The operation ran on spreadsheets and WhatsApp, institutional knowledge lived in two or three heads, and the workshop, with four people for the whole fleet, was flagged internally as a money-loss center. None of it stopped the business. It just taxed it: downtime, slow collections, decisions made without numbers.
Two curves crossed. Inside, a generational transition: new leadership open to better tools, unwilling to bet the operation on a multi-year software project. Outside, AI became good enough to adapt to the company instead of the reverse.
An ERP forces a firm to reshape its processes around software. An AI assistant learns the processes the firm already has: the spreadsheets, the WhatsApp threads, the way the workshop chief describes a fault. For a company that runs on people and paper, that difference is everything.
Instead of selling software, Next2AI embedded an operator inside the company. The method is simple and strict:
Map every process, entity and data source. Find where money actually leaks.
Attack the few processes where AI pays for itself fastest: workshop, receivables, fleet data.
Every system is designed next to the person who will use it. If they don’t adopt it, it doesn’t ship.
Each system carries its own metric from day one.
What works in one unit rolls out to the next: Scania, then Mercedes-Benz, then Volvo.
In traditional companies, adoption beats technology. A modest tool people use every day outperforms a powerful one they ignore.
Diagnostics depended on individual know-how; trucks sat idle and repairs went to outside shops.
Brand-specific AI assistants, starting with Scania, carry unit-level data (VINs, engines, history). A mechanic describes a fault in plain Spanish, on the shop floor, and gets a guided repair.
Receivables spanned entities and currencies, managed by hand. When a key accountant left, the finance function left with them.
AI receivables + cash-flow tooling, a deal-by-deal reconstruction of the Brazil books, and AI-drafted contracts and regulatory briefs.
Decisions on a ~100-unit fleet were made without numbers; master data was incomplete and inconsistent.
A unified ~120-unit master file on one data model, a cloud document hub per business unit, and automated pipelines so numbers roll up instead of being retyped.
Processes lived in people’s heads. Tasks were run by memory and whatever shouted loudest.
A company-wide task and knowledge system with AI-assisted capture, written playbooks, and role-specific copilots for the chief of staff, operations and commercial lead.
High-stakes negotiations ran on instinct, until counterparties arrived with formulas and legal teams.
Every major negotiation gets AI-supported prep: regulatory research, cost-formula analysis, scenario modeling and bilingual Spanish-Portuguese drafting.
“We always knew we weren’t 100% efficient. What we never had was the time or the tools to manage it. Today we have processes and tools that track our operation in real time, and for the first time we manage with numbers, not intuition.”
Five months in, and the rollout continues. The principle stays the same: one process at a time, built with the people who run it, measured from day one.