Skip to content

SELECTED WORK / BUSINESS IN MOTION

Complex problems.
Concrete solutions.

From business needs to delivery validation, see how systems become part of everyday work.
Client identity and production data are hidden. Public interfaces are anonymized redrawings.

THREE COMPLETE REVIEWS

Featured cases are available as focused project reviews.

Each explains the business starting point, first-phase scope, acceptance focus and public boundaries. The full project library remains available on larger screens.

Discuss a similar project
01 · Manufacturing WMS

Stock moves every day. Why do month-end records still fail to reconcile?

A manufacturer's materials, work in progress and finished goods were spread across spreadsheets. The first release addressed three questions before considering a large ERP: where stock is, which batch it belongs to and who recorded each movement.

Anonymous manufacturer / Warehouse & production coordination

Give each material a location, a batch and a traceable movement history.

Background
Excel and chat groups tracked material movements. Finding stock required multiple confirmations, and month-end counts used inconsistent definitions.
Delivery
Material records, batches, locations, receipts, issues, alerts, stock counts, scanning, dashboards and role permissions.
Outcome
Inventory lookup moved from asking colleagues and finding spreadsheets to admin search. Exceptions became traceable, with a shared entry point for material issues and counts.
Acceptance
Key materials can be found by batch and location; movements retain activity records; count discrepancies trace back to business documents.
Dark industrial-style WMS interface for inventory, batches, locations and scanning
Project acceptance evidence Business model Relationships between materials, batches, locations and documents Process records Scanning, receipts, issues and inventory change logs Acceptance checklist Search, traceability, count discrepancies and permission tests Handover documentation Data dictionary, deployment guide and user manual
02 · Mini Program MVP

Enquiries are growing. Why must support still track every booking manually?

Requests came from several channels, with times, locations, services and prices buried in chats. We broke down a complete service into booking, staff acceptance, on-site work and completion confirmation, assigning a clear state to each step.

The client's problem

Many leads, but bookings and service status needed manual follow-up.

Customers arrived through short videos, social feeds and referrals. Support confirmed details in chats, making notifications and follow-ups easy to miss.

What we addressed first

Connect user, staff and admin interfaces first.

Deliver booking forms, time slots, order status, staff acceptance, service records, SMS/WeChat reminders and admin statistics.

Workflow changes

Turn manual progress chasing into traceable work orders.

Booking sources, fulfillment status, staff progress and feedback are centralized, providing a foundation for later membership, coupons and referral features.

Light service-style interface for booking calendars, staff schedules and service progress
Scope baseline Roles, workflows and exceptions Product prototype Key screens and states MVP release Complete booking-to-fulfillment workflow Three connected interfaces Users / Staff / Admin
Project acceptance evidence Workflow baseline Booking, acceptance, fulfillment and completion state diagram Product prototype Critical user, staff and admin journeys Integration validation Notification, duplicate submission and state consistency tests Launch documentation Release list, configuration guide and regression records
03 · B2B Website + FAQ

Your products and technology are strong. Why do customers still struggle to understand your services?

These businesses usually have product information but need to reorganize it around customers. We use common questions to turn complex capabilities into understandable scenarios, case evidence and contact paths.

Blue-and-white editor-style interface for B2B content, leads and conversion data
Understand customer questions first

Turn scattered materials into a website structure.

Extract customer questions, service paths, case structures and FAQs from product sheets, quotes, conversations and previous projects.

Then organize the answers

Launch the website, forms and Q&A materials together.

Deliver the homepage, service pages, cases, downloads, enquiry forms, basic analytics and a knowledge Q&A entry point.

Changes in communication

Sales discussions begin with informed visitors instead of lengthy introductory explanations.

Customers first understand scope and relevant scenarios, then discuss quotes and solutions around their actual needs.

Project acceptance evidence Content sources Customer questions, service boundaries and content structure Search foundations Page titles, structured data and sitemap Conversion journey CTAs, forms, source attribution and lead storage Operations handover Content maintenance, analytics and publishing instructions
04 · Operation Backoffice

Every spreadsheet has an owner. Why is the overall picture still unclear?

The business was established, but customers lived in WeChat, orders in spreadsheets, progress in groups and payments in finance records. The first phase connected one critical workflow.

CRM

Customers & leads

Centralize lead sources, follow-ups, quotes, contracts and owners so opportunities do not remain in personal chats.

Illustrative CRM and sales lead management interface
Order

Orders & projects

Connect quotes, contracts, scheduling, execution, acceptance, invoices and payments through shared statuses.

Illustrative order and project delivery interface
Dashboard

Operations dashboard

Summarize weekly tasks, project risks, receivables, conversion and team work in a view managers can scan.

Illustrative business analytics dashboard
Permission

Permissions & reminders

Provide role-specific access for management, sales, operations, finance and delivery, with due-date, overdue and exception alerts.

Illustrative role permissions and business reminders interface
Project acceptance evidence Workflow rules Lead, order, project and payment status definitions Permission matrix Roles, data scope and action permissions Data migration Field mappings, import validation and exception records Operational reliability Backups, logs, health checks and deployment instructions
What this project taught us

These systems rarely replace every tool in the first release. Begin with a critical workflow such as lead-to-payment or order-to-fulfillment, then connect adjacent processes after launch.

05 · Workflow Assistant

Documents keep growing. Why is the team still answering the same questions?

The problem may be scattered product information, pricing rules and experience rather than staffing. We organize reusable knowledge, then let the system assist with repeated search, summarization and drafting.

Capture experience first

Define what the assistant can answer.

Import product materials, service descriptions, pricing rules, FAQs and after-sales processes, with boundaries for human review.

Then handle repeated work

Let the system handle the first repetitive step.

Generate enquiry summaries, customer profiles, initial replies, proposal outlines, content drafts and follow-up reminders.

Introduce it gradually

Start internally, then expand external access.

Validate accuracy and boundaries with an internal assistant before connecting website Q&A, support channels or business systems.

Dark collaboration-style knowledge retrieval interface with citations and human review
Knowledge inventory Sources, versions and permissions Internal pilot Question sets and failure examples Traceability Q&A and citation records Human review Retain critical review points
Project acceptance evidence Knowledge inventory Document sources, versions and answerable scope Retrieval evaluation Question sets, citation matches and failure examples Human oversight boundaries Review, escalation and sensitive-content rules Audit records Questions, answers, citations, feedback and revision history

Facing a similar problem? We can start with one business workflow.

You do not need a complete requirements document. Share current spreadsheets, conversations and the team's main frustrations so we can identify the priority and define the first release.

Discuss a case solution
Real industry projects · Client information anonymized

From business needs and engineering choices to acceptance evidence, review the delivery process.

The following four examples come from delivered projects. Client identities, locations and production data are hidden; images are redrawn from actual sites and system structures.

Anonymized redraw of an industrial edge gateway and remote maintenance scenario
Anonymized scenario redraw · Client information hidden
Real project 01 · Industrial device maintenance

Add remote status, alerts and diagnostics to existing controllers

The equipment used a proprietary serial protocol and required on-site troubleshooting. The first phase selected one device type to validate parsing, offline backfill, alert work orders and remote logs.

Components
Protocol adapter + Edge gateway + IoT platform
Acceptance
Reconnection, deduplication, alerts and command acknowledgements
Explore smart hardware solutions →
Anonymized redraw of chain-store cold-chain temperature and alert management
Anonymized scenario redraw · Client information hidden
Real project 02 · Retail cold chain

Give headquarters a shared view of store temperatures, exceptions and resolution progress

Stores previously recorded readings manually, detecting issues late and struggling to trace records. The first phase covered selected stores, connecting sensors, thresholds, notifications, acknowledgement and regional escalation.

Components
Sensors + Gateway + Store device platform
Acceptance
Continuous collection, alert escalation and handling records
Explore industry scenarios →
Anonymized redraw of warehouse AI vision receiving and human review
Anonymized scenario redraw · Client information hidden
Real project 03 · AI-assisted receiving

Connect recognition, scanning, human review and stock receipts

Warehouse receiving required repeated package identification and data entry. Edge inference handles recognition, low-confidence results go to human review, and only confirmed results enter the WMS.

Components
Vision workstation + Edge inference + WMS integration
Acceptance
Recognition evaluation, review records and failure examples
Explore AI automation solutions →
Anonymized redraw of an embedded hardware development and testing workstation
Anonymized scenario redraw · Client information hidden
Real project 04 · Smart hardware productization

Turn development-board validation code into testable, upgradable firmware

The product had passed prototype validation but lacked version control, fault recovery and production testing. The first phase fixed the hardware batch and established firmware, communication and testing baselines.

Components
Firmware + Configuration tools + OTA + Production test guide
Acceptance
Stability, fault recovery, upgrades and handover documentation
Explore embedded solutions →

Detailed project reviews