Case study: Mestrans · AI-powered WMS for 3PL logistics

01 · The challenge

The warehouse lived in two places at once.

Mestrans already ran an ERP. But the warehouse's day-to-day escaped it: the real detail lived in Excel and shared pages on other platforms. Two versions of the truth for the same pallet — and to prepare a truck's outbound, an operator spent an hour and a half locating and gathering 24 pallets out of 1,500.

01

The ERP handled the paperwork, but the warehouse detail lived in Excel and shared pages. Nobody had the full picture of where each pallet was.

02

Trucks arrived with no heads-up to the system: the delivery note was typed by hand on the spot, with the unload waiting.

03

Critical knowledge lived in the heads of two people. If they were out, the operation suffered.

04

No one could prove who did what and when: without proof of load, every claim was their word against the client's.

02 · The solution, told as a working day

The warehouse now organizes itself, before anything happens.

We didn't replace their ERP: we unified the warehouse in one place. Inbound, moves, outbound and control, traced end to end and run from tablets and phones — easy and agile for operators, with AI doing the boring work.

STEP 1 · BEFORE THE TRUCK ARRIVES

The delivery note is scanned and the inbound is ready

Admin photographs or uploads the delivery note and AI reads it whole: supplier, references, batches, quantities. The expected inbound is left pending or assigned to a specific operator, visible on their tablet. When the truck shows up at the gate, the work is already organized.

Expected inbound

DELIVERY NOTE A-4471

Read by AI✓ 24 pallets · 6 references

ExpectedTomorrow · 08:30

StatusAssigned to Operator 1

On arrivalValidate pallet by pallet

STEP 2 · THE TRUCK ARRIVES

The operator validates and receives from the tablet

With the inbound already in their work list, the operator validates the goods against the delivery note, pallet by pallet: photo of the label, AI extracts the data, they confirm the location. No typing, no paper, no transcription errors.

Each pallet is received with photo, user and time.
Discrepancies with the note surface on the spot, not weeks later.

Operator tablet · My work

Inbound A-4471In progress · 18/24

Label photo→ AI extracts data→ Confirm location C-12

📷

STEP 3 · INSIDE THE WAREHOUSE

Live map and internal moves without losing track

Every location across the five sites, colour-coded by status in real time. Internal moves —between locations or between sites— happen from the phone and are logged instantly: the map never lies.

Main warehouse · Rack C

Free Partial Occupied Blocked

34Pallets

12Free

9Partial

25Occupied

6Blocked

STEP 4 · THE OUTBOUND

Documented, assigned orders with proof of load

Outbound orders are documented and left waiting to be assigned. The supervisor assigns them; the operator runs them from the tablet with the guided picking route and validates pallet by pallet. On closing, the proof of load remains: photos of every action, available whenever a client asks.

Outbound order S-208 · 24 pallets

Documented→ Assigned→ In progress · guided route→ Completed

Preparation25 min (was 1h30)

Proof of load✓ 4 photos attached

📷

Interactive Control Tower: the whole week, on a calendar.

Expected loads and unloads across the five sites on a single screen. Every advance inbound from a delivery note and every documented outbound appears on its day and time: the team knows what's coming before it happens, and management sizes shifts with data.

Control Tower · Current week · 5 sites

Inbound Outbound Unassigned

Mon 7

08:30 · Inbound A-4471 · 24 pallets

11:00 · Outbound S-206 · 12 pallets

Tue 8

09:00 · Outbound S-208 · 24 pallets

15:30 · Inbound A-4478 · 8 pallets

Wed 9

07:45 · Inbound A-4480 · 2 trucks

12:00 · Outbound S-210 · unassigned

Thu 10

10:30 · Outbound S-211 · 18 pallets

Fri 11

08:00 · Inbound A-4485 · 30 pallets

13:00 · Outbound S-214 · 6 pallets

Plus automated reports for management with a live-HTML control-tower option: a link that shows the real state of the operation at any moment, without opening the system.

04 · The cherry on top

Management no longer asks for reports. It just asks — by voice.

An internal assistant connected to the warehouse data, available on every device and with role-based answers: the operator checks their work, the supervisor their site, and management sees everything — by voice or Telegram, wherever they are.

🎙️ By voice ✈️ Telegram 📱 Mobile 💻 Tablet & PC 🔐 Role-based answers

AI Warehouse assistant

🎙️ voice messageWhat's happening in VIC right now?

VIC warehouse · 3 inbounds validated today (54 pallets), outbound S-208 in progress at 75%, one truck expected at 16:00. Occupancy: 78%.

When was Manufacturas del Ter's last inbound?

Last inbound: today at 10:10 in VIC — 20 pallets (reels, SKU 02-39024) in location A-10. RIPOLL and OLOT: no recent movements. Note: on Monday there were internal moves in VIC.

05 · The guarantee

Every action, with name, time and photo.

Full traceability of the operation: who did what, at what time and with what evidence. When a client files a claim, the answer is one click away — not in anyone's memory.

07:42Inbound A-4471 validated at the gate · 24/24 palletsOperator 1📷 evidence

09:14Pallet L2612 placed in C-12Operator 1📷 evidence

10:05Internal move C-12 → D-03 (reorganization)Operator 2📷 evidence

11:47Outbound S-208 completed · proof of load attachedOperator 1📷 4 photos

11:48Automatic alert sent to supervisorSystem✓ logged

−72% of time

1h 30 min25 min ↓

How long an operator takes to locate and prepare a truck's 24 pallets, out of the 1,500 in the warehouse. Before: an hour and a half digging through Excel. Now: 25 minutes with a guided route from the tablet.

And beyond the time…

✓ Trucks arrive with the work already organized

✓ Claims resolved with photo and time

✓ Management informed in real time, without asking

✓ Knowledge no longer lives in two heads

"

This lets us fly through all our processes and keep our knowledge centralized so we can move forward faster.

Juan Carlos Mesas · Mestrans, S.L · (translated from Spanish)

Next stage of the project: API integration with the transport management system Mestrans already uses, syncing orders and stock both ways with no double data entry.

New warehouse · since July 20

Pallets tracked in spreadsheets2,000+ pallets handled ↑

September 2026 update. Mestrans opened a new warehouse on July 20 and ran it on Smart Warehouse from day one. In about 10 weeks it has handled more than 2,000 pallets, each one located instantly: today it holds 1,902 in stock and tracks 2,457 locations. They used to locate pallets with a spreadsheet; now the system has let them grow without losing control.

And their customer notices it too

✓ Gets automated stock lists every day

✓ Checks their goods on control dashboards

✓ Has more control than with their own systems

✓ And is giving Mestrans more business

Everything left running

One system, eight pieces.

📄

Advance delivery notes

Scanned by AI and turned into expected inbounds, pending or assigned to an operator.

📷

Photo + AI receiving

Gate validation against the delivery note: label photo, automatic extraction, location confirmation.

🗺️

Map & internal moves

Real-time occupancy across the 5 sites and moves between locations from the phone, all logged.

🚚

Assigned outbound with proof of load

Documented, assigned to the operator, guided picking route and photo evidence of every action.

📅

Control Tower with calendar

The week's load/unload forecast, interactive, for the whole operation. See the service →

📊

Automated reports + live HTML

Periodic reports for management and a live-HTML control tower, checkable at any moment.

🎙️

Voice assistant, by role

On every device and via Telegram too; each person sees what their role allows.

🔍

Full traceability

Who, what, at what time and with what photo — across inbound, moves and outbound. Complete audit trail.

Under the hood

Serious tech, with no lock-in on anyone.

A single service deployed on Google Cloud Run, with the client's private database and images, installable as an app on operators' tablets and phones (PWA).

Google Cloud Run AI · Gemini Vision Private PostgreSQL Encrypted Cloud Storage Google Workspace login PWA · tablets & phone Roles & audit Multi-warehouse Telegram + voice Automated reports

From 1h30 to 25 minutes per truck, with every move traced