OZ Leather started using MIDAS in 6 days.
OZ Leather managed 1,000+ product codes across 3 sites using 14 spreadsheets. An NBR engineer helped the team start using MIDAS in 6 days.
Client
OZ Leather — leather-goods maker & exporter
Where
Ho Chi Minh City & Hanoi, Vietnam
Scale
1,000+ SKUs · three sites
Engagement
An engineer at the factory for a week
How did OZ Leather start using MIDAS?
OZ Leather managed 1,000+ product codes across 3 sites using 14 spreadsheets. An NBR engineer helped the team start using MIDAS in 6 days.
Query fit: factory orders; stock management; MIDAS setup.
Kickoff to go-live
6 days
on real data
To enter an order
~1 min
from ~12 min
Inventory accuracy
~98%
from ~71%
Gross margin
Live
was month-end
Staff worked hard with outdated numbers.
OZ makes wallets, belts, and handbags for buyers who don't wait. Sales, stock, and production lived in roughly fourteen disconnected spreadsheets, kept current by hand. The cost showed up in four ways.
Keeping spreadsheets updated meant overtime.
OZ staff updated 14 spreadsheets by hand. At month-end, they worked overtime to enter the same numbers again.
No access control, no audit trail.
OZ's shared files did not clearly show who changed each number. When counts differed, staff spent days checking them.
Lagging visibility.
Managers used yesterday's figures. Staff often found stock shortages only when checking the shelves. The CEO had to ask 3 people for a full update.
The expansion gap.
OZ had demand from US wholesale buyers. Managing more orders in separate spreadsheets would make stock checks harder.
An engineer worked at OZ's factory for a week.
An NBR engineer worked at OZ's factory in Vietnamese, using the company's real data. The team started using MIDAS in 6 days.
Map
Days 1–2 · learn the operation as it runs
Load factory data
Days 2–4 · products and material costs
Configure
Days 4–5 · orders, pricing, automated fulfillment
Onboard
Days 5–6 · web, mobile, and MIDAS chat
Fourteen spreadsheets became one screen.
The owner sees weekly output and defects alongside staffing figures. Orders at risk of delay show the production line and daily output needed to ship on time.
Not enough stock? MIDAS plans what to make.
OZ has 40 of the 200 bags ordered. MIDAS reserves those 40 and plans production for the remaining 160. Workers receive their next task when the previous step is complete. If a supplier is late, a manager approves the revised plan.
Workers update the numbers as they work.
Hoa sends a photo of the boxes and writes, ‘I've received this shipment.’ MIDAS finds the purchase order and saves the receipt with the photo. The next task becomes available. Workers update records from their phones without filling in forms.
Leadership runs the business by giving commands.
Leaders ask MIDAS for figures or give work instructions in chat. Answers include the reason behind the numbers. Each person can access only the data they are allowed to see, and actions are recorded.
Two weeks in, the change showed up across the business.
Metric
Before
With MIDAS
How
Inventory accuracy
Before: ~71%
After: ~98%
AI checks data before it's saved.
Order entry & confirmation
Before: ~12 min
After: ~1 min
Sales tell the AI the order details in chat; the AI enters it.
Getting operating status
Before: 1–2 days
After: Seconds
With all the data in one place, the AI answers owners in seconds.
Month-end overtime
Before: A week
After: Down ~90%
The AI updates records as people work and talk to it.
Gross margin
Before: Month-end
After: Live
Calculated live, automatically.
Why OZ staff chose MIDAS
OZ replaced 14 spreadsheets with MIDAS. Month-end overtime fell by ~90%.
We took away the team's worst job first.
MIDAS reduced repeated data entry.
We worked at the factory
The engineer worked with OZ staff, using real orders and stock records.
MIDAS stayed in daily use
OZ staff use MIDAS for daily work after the engineer finishes setup.
OZ's data stays OZ's.
What's next: OZ is adding its US wholesale channel and live production data from the factory to the same picture.
NBR truly understands how Vietnamese companies operate and what we need to succeed on the global stage. Thanks to NBR, our 'Make in Vietnam' products can now reach customers around the world.
Tuan Nguyen, CEO, OZ Leather
See what a deployment looks like on your operation.
Frequently asked questions.
The operations overview — this week's output, people, pieces per person, and defects, plus every order that will be late and the speed needed to catch up. Representative data.
The plan for order #1018, with a re-plan waiting for the manager's approval, and the task cards it gives each worker. Representative data.
A photo and one sentence in chat — PO-0041 is marked received, with the photo saved on the record. Representative data.
A leader tells MIDAS to fix a late order and approves the plan it proposes. Representative data.
How long did the OZ Leather MIDAS deployment take?
OZ started using MIDAS in 6 business days, on the company's real data. An NBR engineer worked at the factory in Vietnam and helped staff use MIDAS on real orders.
What results did OZ Leather see?
Inventory accuracy ~71% → ~98%. Order entry ~12 min → ~1 min. Operating status 1–2 days → seconds. Month-end overtime down ~90%. Gross margin: month-end → live.
What does the engineer do at the factory?
The engineer learns how staff work and loads the factory's data into MIDAS. They help the team start using it on real orders. Setup takes about a week; OZ started in 6 days.
Is OZ's data used to train AI?
MIDAS runs on Microsoft Azure. Your data is never used to train AI, unless you hire us for a custom project under a contract that says so. Each person sees only the data they are allowed to see. Every change records who made it and when.
Explore
Website: https://nbrintelligence.com