Zeror
EN中文
Tell Us About the Work
Menu

Cases

Real Problems.Real Systems.Real Numbers.

Every case here comes from one working business: a store, a warehouse, a 64,000-product online catalog, dozens of suppliers. Each starts with a problem the team hit every day. Each ends with a result, and says exactly how it was measured.

7 client cases from one working business, plus 2 other projects.

Each card reads top to bottom: the problem, what we built, and the result.

Case 01online catalog

Measured

Keeping a 64,000-product online catalog accurate

Problem
Supplier prices, images and stock kept drifting out of sync with a 64,000-product website.
What we built
One controlled update pipeline: match, price diagnostic, human approval, bulk write with rollback, read-back check.
Result
386 / 386price updates verified correct after writing

Read the case

Case 02POS migration

In progress

Moving from a legacy POS to Square

Problem
The old POS fed the whole back office: order follow-up, delivery, purchasing and commissions.
What we built
Migration to Square with sync back into the existing system, automatic ID-conflict handling and a parallel test site.
Result
66,000+products in the old POS; tooling built and dry-run

Read the case

Case 03legacy software

Measured

Automating data entry into legacy desktop software

Problem
Product and receiving data were typed by hand into desktop software that could not be replaced.
What we built
An automation layer around it: structured input, desktop automation, logs and human checks.
Result
21,531product records entered without manual typing

Read the case

Case 04orders & purchasing

Operating estimate

Reconnecting sales and purchasing

Problem
The same order information was re-entered and reorganized across sales and purchasing.
What we built
Orders captured once, structured, and passed from sales to purchasing.
Result
≈ 67.7 hof manual work removed each month

Read the case

Case 05price & stock lookup

Operating estimate

One lookup for prices and inventory

Problem
Staff searched supplier price files, internal stock and supplier stock separately.
What we built
One lookup across all sources, with model matching and pricing calculations.
Result
≈ 40,000products in one lookup; 2,000+ lookups a month

Read the case

Case 06showroom

Operating estimate

Tracking showroom samples

Problem
Finding a showroom sample, and whether it was installed, depended on memory and walking around.
What we built
Shared tracking of 990 samples by location, installation status and history.
Result
5 min → 30 sto find a sample

Read the case

Case 07supplier documents

Tested

Turning supplier documents into structured product data

Problem
Invoices, price lists and spec sheets arrived as PDFs and spreadsheets in every format, and were re-keyed and cross-checked by hand.
What we built
Rule-based extraction and matching: invoice PDF to a fixed-format sheet with a reconciliation total, model matching that flags ambiguity, specs filled from official sources.
Result
44 / 45models matched in a trial against ~6,500 records; 10 ambiguous matches sent to a person

Read the case

How to read the results

Measured
Counted from logs, read-back checks or data snapshots.
Operating estimate
Usage frequency × time per task, confirmed with the team.
Tested
Run on real business data and reviewed, but not yet in daily production.
In progress
The project is still under way; results are not final.

Other projects

Client project · interior visualization

From CAD to client-approved interior renders

A commercial storefront interior built from the client’s CAD and design, with one color standard across every view and several revision rounds. The client approved the final renders.

Which task takes too much time?

Tell us how your team does it today and which software and files are involved. We can work through where a tool or a different process would help.

Discuss Your Project