Skip to content
MK
Menu
中Contact
M.K.Products and AI WorkflowsTaiwan × Canada

From a market need to a working AI product.

I work across Taiwan and Canada on AI products, content systems and e-commerce workflows. This site documents current projects, working methods, and my writing on future cities and technology in society.

Product development · AI workflows · e-commerce operations · market validation

M.K.
M.K. / VancouverTaiwan × Canada
Selected projects

Projects in progress, from early demand signals to working products and delivery processes.

The cases below cover AI tools, websites, real-estate research, fashion imagery, e-commerce operations and brand work. Each one shows the problem, the current approach and the signals I use to decide what to change next.

01In active testing
2025–2026Product and workflow designBusiness development, CRM, early market validation

AI Auto BD Kits

A reusable outbound workflow for testing early products and small-business offers.

AI Auto BD Kits connects audience definition, lead preparation, Google Sheets CRM, Gmail API outreach, reply syncing and weekly review. Its purpose is not to increase email volume, but to record why people reply or decline so the next audience, message and offer can be revised.

Problem

An early product may have a working prototype without a clear view of whom to approach, how to explain its value or how to interpret market response.

Current approach

Lead sources, CRM fields, outreach cadence, reply labels and weekly reviews form an adjustable business-development workflow.

What I watch

Reply rates, qualified conversations, reasons for rejection and changes in the next cycle show whether further work is justified.

Current outputs

Google Sheets CRMGmail API outreachLead-cleaning rulesWeekly review notes
02Pilot and workflow validation
2025–2026Workflow design and production QAFashion e-commerce, virtual try-on, image operations

VTO Ultra

An AI virtual try-on and batch image-production workflow for fashion e-commerce.

VTO Ultra covers the asset standards, consistent models, service providers, quality checks and cloud delivery required for high-volume SKU imagery. The question is not whether one image can look convincing, but whether the same process can handle volume, garment variation, rework and brand consistency.

Problem

A single AI fashion image can make a strong demo; volume, fit variation, consistency and delivery quality are the harder constraints.

Current approach

Asset specifications, model rules, provider tests, QA criteria, batch naming and cloud delivery define the workflow.

What I watch

Pass rate, rework rate, delivery time and brand feedback indicate whether the process is stable enough to repeat.

Current outputs

SKU production workflowModel-consistency rulesProvider comparisonQA checklist
03Product testing
2025–2026Product strategy and report workflowCanadian real estate, Chinese-language market, research content

PropertyLens / InsightEstate

Real-estate research reports for Chinese-speaking buyers and agents in Canada.

PropertyLens / InsightEstate organizes neighbourhood data, property information, buyer questions and risk factors into readable research reports. The reports are intended to help Chinese-speaking readers understand Canadian real estate and to give agents a more complete basis for explanation and follow-up.

Problem

Chinese-speaking buyers need information that connects local rules with their own decision context, while agents need a format more substantial than a social post.

Current approach

Area research, property interpretation, buyer questions, report templates and delivery methods form a repeatable research process.

What I watch

Reading, forwarding, follow-up questions and consultation requests indicate whether the report is improving understanding and trust.

Current outputs

Research report templateBuyer decision frameworkChinese-language positioningAgent workflow
04Active
2024–2026Founder and project deliveryAI adoption, content, websites, business workflows

FlyPig AI

A consulting and delivery practice for applying AI to business workflows.

FlyPig AI works on content, websites, business development, search visibility, e-commerce and enterprise adoption. Each project begins with the existing workflow and its constraints, then identifies where AI is useful and which parts should become tools, templates or documented delivery steps.

Problem

Many teams want to try AI but are unsure which task to begin with or how to turn scattered tools into a process people can keep using.

Current approach

Client work and owned brands are used to test approaches across content, websites, business development, delivery and internal operations.

What I watch

Ongoing use, reduced repetitive work and the team's ability to understand and maintain the process show whether an approach is useful.

Current outputs

AI workflow templatesClient delivery notesContent systemSEO / GEO / AEO tests
05Market testing
2025–2026Product, marketing and delivery designLocal businesses, SEO, GEO, website deployment

3Kup SEO

An AI-assisted website and search-content service for local businesses.

3Kup SEO defines a clear service scope around website production, search structure, FAQs, Cloudflare deployment and enquiry forms. It begins with needs a local business already understands—being found, understood and contacted—rather than asking the client to learn AI terminology first.

Problem

Local businesses need online visibility but may not have time for complex technical explanations or a large consulting engagement.

Current approach

Pages, FAQs, search intent, enquiry forms and deployment steps are organized into a defined delivery package.

What I watch

Enquiries, form submissions, search coverage, client comprehension and delivery time show whether the service is appropriately scoped.

Current outputs

Website delivery workflowSEO / GEO FAQ templatesCloudflare Pages deploymentEnquiry-form routing
06Active operation
2020–2026Brand and e-commerce operationsE-commerce, outdoor products, day-to-day operations

ICareU / Horizon Outdoor

Active e-commerce and outdoor-brand operations used as a practical testing ground.

ICareU and Horizon Outdoor provide real constraints—products, inventory, customer service, sales and platform rules—for testing AI in product copy, advertising assets, content management, order organization and daily operations. This keeps other workflow experiments close to the conditions they are meant to support.

Problem

Without a real operating environment, it is difficult to know whether an AI tool can handle ordinary, repetitive work that still affects the business.

Current approach

Product data, content production, advertising assets, order processes and brand operations provide a continuing test environment.

What I watch

Production time, content consistency, process errors, support workload and sales feedback show whether AI is reducing friction.

Current outputs

E-commerce operationsBrand content workflowAdvertising testsOrder and inventory notes
Approach

My work usually begins with a specific problem.

I first identify who is affected, how the problem is handled today and what outcome is worth testing. Only then do I decide whether AI, a website, automation or another tool belongs in the solution.

01

Clarify the problem and user

Identify who is already spending time, money or attention on the problem, then describe the present constraints as a testable assumption.

02

Build a usable first version

Make enough for a person to use or respond to. The first version does not need every feature, but it must move beyond a presentation.

03

Use real responses to revise it

Test the offer through content, outreach, search, interviews or partnerships, and record what needs to change in the audience, message or delivery.

04

Document what proves useful

When a method works repeatedly, document the data, steps, responsibilities and review points so the next cycle does not start from zero.

Working workflows

A method becomes a workflow only after it has been tested.

These workflows are still in use and under revision. They keep records of leads, content, data, delivery and review so repetitive work can be reduced and changes can be made with evidence.

01

AI Auto BD Kits

Audience / leads / CRM / outreach / weekly review

Breaks business development into trackable steps so replies and rejections can inform the next audience, message and offer.

02

3Kup SEO

Website / search structure / FAQ / forms / deployment

Defines a clear delivery scope for local-business websites, search content, enquiry paths and ongoing maintenance.

03

VTO Ultra

SKU assets / virtual try-on / QA / delivery

Moves fashion imagery from isolated generations into a production workflow with standards, review, rework and delivery.

04

PropertyLens

Area research / buyer questions / reports / agent use

Turns scattered real-estate information and buyer questions into reports that explain properties, neighbourhoods and risks.

Related projects

Different projects test the same working methods under different constraints.

E-commerce, real estate, content and visual production each create different limits. These sites show how the work changes from one context to another.

01
EC Ultra
02
Elite Fashion
03
Neo Babylon
04
IP Verse AI
Notes and essays

I am more interested in what changes when AI enters real work.

These notes cover open questions in products, education, work and society. Some positions will change as the evidence and my own experience change.

Note

Many AI projects stall outside the model

A capable model is not enough when data, ownership, workflow and adoption have not been made clear.

Note

Market validation starts before the demo is finished

Testing who will use a product, why it matters now and how it will be delivered can prevent a weak direction from becoming an elaborate prototype.

Note

Small-business AI adoption should begin with existing work

Most businesses do not need another tool list. They need a practical path from the way work happens today to a process they can evaluate and maintain.

Contact

Start with a specific problem.

If you are working on an AI product, an e-commerce or content workflow, a Taiwan–Canada market question, or future-city and science-fiction work, send a short description of the situation and what you would like to discuss.

Helpful context: what you are building, what is currently blocked, and how you would like me to participate.

Email M.K.FacebookView projects