CASE STUDIES / EVIDENCE
Show how
decisions changed.
We show who made each decision before implementation, which information and rules moved into software, what remained human, and which proof is public or still undisclosed.
- ImplementedOnly functions we actually built
- Public scope statedNo private numbers converted into claims
- Human decisions shownBoundaries that stay manual are explicit
CASE EVIDENCE MAPImplementation / observation / next validation
Cases6Reframed on this page
Quant effectUndisclosedUndisclosed stays undisclosed
ExtensionsSeparateUnimplemented ideas are marked
EDITORIAL RULE
We do not mix what was built with what could be built next. Cases without deployed optimization are treated as related implementation proof.
SCROLL TO EVIDENCE
HOW TO READ
Read each case
in three layers.
Implemented, observed, and next automation candidate are different claims. This page separates the labels and keeps every claim within the available proof.
01
IMPLEMENTED
Published implementation
Functions, period, and technology confirmed by public records or approved evidence.
Published as fact02
OBSERVED
Confirmed workflow change
Changes like searchable schedules or mobile approvals, without estimating unpublished savings.
Within public scope03
NEXT HYPOTHESIS
Future mathematical automation
Ideas that can be tested with existing data, always marked as not yet implemented.
Proposal, not implementedWhat we do not show
We do not show unsupported reduction rates, unapproved customer details, unimplemented optimization as delivered work, or claims such as fully automatic unless the evidence supports them.
CASE NAVIGATOR
Find cases close to your work
with three questions.
The navigator uses decision type, users, and evidence needs, not just industry names.
STEP 01 / DECISION
What decision is mostly done by people?
Choose the closest one.
STEP 02 / USERS
Who shares the information?
Choose the main users.
STEP 03 / EVIDENCE
What do you want to confirm?
This weights the recommendation reason.
Recommended case
01
Matching / scheduling
Home-visit nursing schedule management
A system that lets care managers check and manage home-visit nurse schedules in real time.
FEATURED CASE 01
Home-care visit schedules,
checked and managed in real time.
This is not a deployed mathematical optimization case. It is the closest public case to the current focus because it handles visit, staff, and time data and implements the workflow for care managers to check and manage schedules.
BEFORE / IMPLEMENTED FLOW
BEFOREPre-implementation details are not public
Before processNo public detail
The public record does not state the previous checking method, number of contacts, or time required. We do not fill those gaps with estimates.
IMPLEMENTEDReal-time schedule checking and management
Care managerSystemVisit scheduleCheck and manage
We implemented a system that lets care managers check and manage home-care nurse visit schedules in real time.
Implemented scope
- Nurse schedule display
- Care manager checking screen
- Real-time visit schedule management
- Mobile-friendly screen
Decisions kept human
- Final clinical fit check
- Agreement with users and families
- Response to urgent changes and exceptions
- Formal request and confirmation decisions
NEXT HYPOTHESIS / NOT IMPLEMENTED
Mathematical automation to validate next
Qualification constraintsVisit time windowsTravel timeContinuity of staffFair workloadAutomatically create assignment candidates
View visit scheduling solution
CASE LIBRARY
Explore public work
by decision type.
Home-visit nursing schedule management
A system that lets care managers check and manage home-visit nurse schedules in real time.
Global order management system
A member-based order system with English UI for overseas buyers, product management, customer management, messaging, documents, and catalogs.
Self-pick vegetable matching app
An app for matching producers and consumers so people can pick vegetables directly in the field and purchase products.
Timecard management for construction sites
A system for construction staff who go directly to sites to clock in and out and request paid leave from a smartphone.
Visual English dictionary app
A dictionary app for learning English in more than 30 languages, implemented with SQLite, Flutter, and Firebase.
Translation support tool Isometry
A web system that supports fast, high-quality translation. The public record says it was adopted in a Deloitte Tohmatsu translation project.
No cases match these conditions. Shorten the search term or choose All.
MORE IMPLEMENTATION PROOF
View demos You can also review implementation quality in the demo library.
Read-only system demos for logistics, manufacturing, retail, sales, construction, and more are collected on a separate page.
FROM SYSTEM TO OPTIMIZATION
Connect existing implementation
to the next calculation.
These are not delivered functions. They are future hypotheses that should first be checked through a small prototype.
Implemented
Visit schedule visibility
People, time, schedules, external users
Not implemented / validation candidate
Automatic assignment of staff to visits
Evaluate qualifications, time windows, travel, and continuity together
View related pageImplemented
Unified orders and documents
Products, orders, customers, progress, documents
Not implemented / validation candidate
Processing, inventory, and shipping priority
Create candidates from due dates, stock, margin, and logistics conditions
View validation methodImplemented
Time tracking, requests, approvals
Staff, attendance, leave, approvals
Not implemented / validation candidate
Shift plans from preferences and required staffing
Create candidates that respect skills, consecutive work, and fairness
View related pageImplemented
Product publishing, inventory, payments
Producers, products, buyers, inventory
Not implemented / validation candidate
Supply and demand matching candidates
Generate recommendations from area, stock, freshness, and preferences
View related pageCASE DETAIL STANDARD
Publish every detailed case
with the same eight fields.
Instead of listing only convenient numbers case by case, we use a common comparable format.
- 01Business background
Who was struggling, and in which situation
- 02Before decision
Public information and confirmed prior decisions
- 03Data used
People, time, products, schedules, permissions
- 04Implemented scope
Screens, data, notifications, integrations
- 05Human decisions
Exceptions, approvals, expert judgment boundaries
- 06Observed change
Workflow and usage changes
- 07Public proof
Period, technology, links, and whether numbers exist
- 08Next hypothesis
Unimplemented mathematical automation in a separate field
FAQ
Frequently asked questions
about case studies.
Why do some cases not show reduction rates or sales growth?
Because no public or approved measurement is available. We show implementation scope, period, technology, and workflow change without inventing metrics.
Do all cases use mathematical optimization?
No. The page includes related implementations such as data foundations, schedule management, matching, permissions, search, and language tools. Cases without optimization are labeled that way.
Can we ask even if there is no case in our industry?
Yes. We compare decision structures such as assignment, ordering, scheduling, approval, and matching rather than only industry names.
Can a case be published without the customer name?
Yes. Industry, scale, problem, and scope can be anonymized, then checked before publication.
Can we validate with our own Excel data?
Yes. We first check columns, volume, hard conditions, preferences, and current decision steps, then test with a small prototype.
Your next case
The next case
could be your operation.
A same-industry case is not required. We start from your current spreadsheets and recurring decision conditions to identify what can be automated.
What we check first
- 01Current spreadsheets and forms
- 02Hard constraints and preferred conditions
- 03Metrics to compare against the current plan