Orfeu's distribution center became the bottleneck before its warehouse management system became the fix
A year ago, the distribution center for Brazilian specialty coffee brand Orfeu handled 100 to 150 e-commerce orders a day, with much of the control still done on paper. After deploying a warehouse management system (WMS) and handheld data collectors, that same center has already processed 800 orders in a single day [1].
The case was presented on October 13 during Universo Totvs 2026 in Sao Paulo, alongside Grupo BT, a distributor of personal protective equipment and maintenance supplies that ran its own WMS rollout [1]. Both companies used the panel to make the same point to the room: the gap between front-end sales growth and what a warehouse can actually execute is not closed by selling more.
It is closed by knowing exactly where a product sits, how long a picker takes to build an order, and how much a manager can see without walking the floor [1].
Why are most warehouses stuck below the level where AI pays off?
Gartner's five-level intralogistics maturity scale, cited during the panel, ranks warehouse operations from manual to autonomous, and places 72% of companies at level three or below, short of the point where artificial intelligence starts anticipating problems instead of just recording them [1].
The five levels, as presented at the event, are:
Totvs's own view is that reaching level four requires genuine end-to-end data on a single platform, not systems that merely pass an invoice or a customer record from one tool to another. The company said it spent three years just defining that data architecture for its logistics products [1].
That timeline matches what Gartner itself has been telling supply chain leaders more broadly: warehousing has reached an inflection point for AI, driven by labor constraints that make automation non-negotiable, capital models that lower the cost of entry, and AI and autonomy technologies that have reached operational maturity [2].
- Manual
- Controlled
- Connected
- Intelligent
- Autonomous
Fixing the data came before adding any automation
Before Orfeu could automate anything, it had to clean up a database built around years of manual habits. Ricardo Madureira, the company's CEO, said the pressure came directly from the e-commerce customer, who expects a tracking code fast, which exposed how disorganized the underlying data had become [1].
“We had to do a massive cleanup of our database, because we were used to a manual process, and that is where we saw the opportunity to get the house in order.”
Once the data collectors were in place, the team started generating and fulfilling orders within carrier cutoff windows, and operational indicators moved onto televisions placed around the warehouse. Operators gained direct access to information like empty storage slots and stock balances, cutting their dependence on a supervisor and giving them more autonomy over daily decisions [1].
Grupo BT moved inventory before it moved software
At Grupo BT, the WMS went live six months before the panel. Itamar Rogerio Bacho, the company's IT and systems manager, said the change started with management itself: the logistics manager used to work the floor during operations, and now follows everything from a dashboard [1].
Using a heat map of the warehouse, the company relocated its highest-turnover products closer to the conference area and to the lower levels of its five-tier shelving, so pickers no longer had to wait for a forklift to reach a fast-moving item [1].
| Metric | Before WMS | After WMS |
|---|---|---|
| Orfeu daily e-commerce orders | 100-150 [1] | Up to 800 [1] |
| Grupo BT orders separated per shift | Baseline | +20% [1] |
| Grupo BT logistics manager's role | Working the floor | Managing from a dashboard [1] |
Can traceability turn warehouse data into an answer customers can use?
When a Grupo BT customer complains about a missing or defective item, the company can now check exactly when and how that box was verified, turning a warehouse log into a direct answer for a support ticket [1]. That same data changed how the commercial team operates.
Before the WMS, sales staff sometimes promised stock that was not actually on the shelf. Bacho described the result plainly: the customer would arrive at the shelf position and find nothing there, an unserved order with no warning [1]. Today the commercial team trusts the stock figure the system shows.
Training time dropped too. The conference team, now staffed mostly by women through an inclusion program at the company, learns to operate the system in one or two days [1].
AI already turned a 30-click lookup into one voice message
One year into its WMS project, Orfeu opened AI use to its operational team. Three staff members who knew the warehouse routine well started using Totvs's AI assistant to automate tasks and build their own reference guides [1].
The before-and-after is concrete: tracing an item's movement through the management system used to take around 30 clicks. Now, an operator sends a voice message and gets the traceability answer back immediately [1].
Totvs is also building agents still in development, including one for receiving docks that indicates which invoice to check first when several trucks arrive at once. The agent bases that call on which stock is running below target and which pending orders are waiting on it, and it suggests the fastest available employee for the job [1].
Evandro, Totvs's data product lead, framed the goal as augmentation, not replacement: the point is to improve human capacity and business capacity together, not to substitute one for the other [1].
Why a coffee brand's warehouse case matters for peak season planning
The sequence Orfeu and Grupo BT describe, clean data first, visibility second, AI assistance third, mirrors what Gartner identifies industry-wide as the actual blocker to scaling warehouse automation [1][2].
Commerce teams preparing for a holiday season that keeps compressing delivery expectations face the same ordering problem these two companies solved internally. Warehouse automation has grown from a niche investment into a crowded market: Statista counts hundreds of dedicated warehouse automation companies operating worldwide, with the largest concentration in autonomous mobile robots and automated guided vehicles [3]. The technology options exist.
Sources
- [1]E-Commerce Brasil, Como a Orfeu saiu do papel e chegou a 800 pedidos por dia no e-commerce — ecommercebrasil.com.br
- [2]Gartner, Gartner Identifies the Top Four AI Trends Transforming Warehousing for Supply Chain Leaders — gartner.com
- [3]Statista, Supply Chain — statista.com
