M.K.Product / Workflow / GTMTaiwan - Canada

AI projects, from unclear signals to working systems.

M.K. works across product planning, workflow design, content architecture, business models and business development, helping early AI projects move from concept to prototype, from prototype to market, and from market feedback into repeatable operating systems.

View selected work Contact

Portrait / Field noteAI work as operating practice

Selected Work

The work is not a resume list. It is a record of methods placed into live markets.

These projects span outbound systems, SEO/GEO, real estate research, AI image production, e-commerce operations and brand experiments. Each case keeps the problem, system and signal visible, because a useful project should be able to inform the next one.

2025-2026Venture Builder / Workflow ArchitectOutbound, CRM, SMB GTM

AI Auto BD Kits

A reusable outbound system for early-stage AI and SMB offers.

AI Auto BD Kits turns markets, regions, languages and customer segments into a repeatable outbound workflow: lead sourcing, deduplication, Google Sheet CRM, Gmail API sending, reply sync and weekly iteration. The point is not to send more email. The point is to help early products test market signals before the offer is fully fixed, and to let each reply, rejection and conversation feed back into the next version of the offer.

Problem

Early AI products often have a demo, but not a clear view of whom to contact, how to frame the offer, how to record replies, or when to judge that a market signal is not strong enough.

System

Lead sources, CRM fields, send cadence, reply labels and weekly reviews form a reusable outbound loop.

Signal

Reply rate, qualified conversations, rejection reasons and offer changes indicate whether a market is worth another cycle.

Google Sheet CRMGmail API workflowLead cleaning rulesWeekly iteration notes

2025-2026Product / GTM / Delivery SystemLocal business, SEO, GEO, Cloudflare

3Kup SEO

A low-friction AI-assisted SEO / GEO website offer for local businesses.

3Kup SEO packages websites, SEO, GEO FAQ, Cloudflare deployment and lead forms into a service local businesses can understand and adopt with low friction. It does not sell AI as an abstract vision. It places AI inside familiar business needs: being found, being compared and being contacted. The project tests how AI capability can enter SMB markets through a clear offer and a delivery process that reduces education cost.

Problem

Local businesses need online visibility, but often do not want to start with complex technical explanations or large consulting retainers.

System

Site generation, FAQ architecture, search intent, lead forms and deployment steps are turned into a fixed delivery package.

Signal

Inquiries, form submissions, keyword coverage, customer comprehension and delivery cycle time show whether the offer is working.

Website delivery workflowSEO / GEO FAQ templatesCloudflare Pages deploymentLead form routing

2025-2026Product Strategy / Report WorkflowReal estate, Chinese market, research content

PropertyLens / InsightEstate

AI real estate research reports for Canadian Chinese buyers and agents.

PropertyLens / InsightEstate combines Canadian real estate information, buyer decision-making, immigration context and agent expertise into AI-assisted research reports. The work is not about one-off content. It is about helping agents use structured research and narrative to make locations, properties and risks easier to understand, while turning the report into a trackable commercial touchpoint.

Problem

Chinese-speaking buyers need real estate information that understands context, while agents need a more trust-building format than ordinary social posts.

System

Area research, property interpretation, buyer questions, report templates and delivery channels form a repeatable content workflow.

Signal

Report opens, forwarding, consultation questions and mandate intent show whether the content is creating trust.

Report templateBuyer decision frameworkChinese-language positioningAgent-facing workflow

2025-2026Workflow Design / Production QAFashion e-commerce, VTO, image operations

Fashion Ultra

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

Fashion Ultra focuses on high-volume SKU image production, input standardization, fixed AI models, VTO providers, QA and cloud delivery. The central question is not whether a single generated image can look impressive. It is whether AI image production can move into a controlled production rhythm with consistent inputs, reviewable quality, recoverable errors and useful delivery for brand or e-commerce teams.

Problem

AI fashion imagery demos can look strong, but volume, fit variation, brand consistency and delivery quality are the harder constraints.

System

Asset standards, model rules, provider testing, QA criteria, batch naming and cloud delivery define the production workflow.

Signal

Pass rate, rework rate, delivery time, brand acceptance and repeatability indicate whether the workflow is stable.

SKU production workflowModel consistency rulesVTO provider comparisonQA checklist

2024-2026Founder / OperatorAI adoption, content, websites, business workflows

FlyPig AI

A field lab for turning AI tools into business workflows.

FlyPig AI is the brand M.K. uses to experiment with and deliver practical AI business adoption. It spans content, websites, business development, SEO/GEO/AEO, e-commerce workflows and enterprise adoption. The working style is pragmatic: place the problem inside a workflow, decide where AI should enter, then turn repeatable parts into tools, templates or delivery systems.

Problem

Companies and founders often know AI matters, but do not know which workflow to begin with or how to turn tool trials into operating capability.

System

Content, websites, BD, delivery and internal workflow projects accumulate reusable AI adoption patterns.

Signal

Real clients, owned brands and market tests show which methods can keep working after the demo stage.

AI workflow libraryClient delivery patternsContent systemSEO / GEO / AEO experiments

2020-2026Operator / Brand BuilderE-commerce, outdoor brand, operations

ICareU / Horizon Outdoor

An operating ground for e-commerce, brand and AI workflow experiments.

ICareU / Horizon Outdoor provides real e-commerce and outdoor brand operating contexts for testing AI in product copy, ad assets, content systems, order organization and day-to-day workflows. These experiments are not staged in imaginary conditions. They face products, inventory, customer service, sales and platform constraints, keeping AI workflow design close to the work it is supposed to support.

Problem

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

System

Product data, content production, ad assets, order processes and brand operations become a working test ground.

Signal

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

E-commerce operationsBrand content workflowAd creative experimentsOrder and inventory process notes

Praxis

Moving AI projects through problems, workflows, tools, sales and iteration.

Praxis here is a working order. Understand how the problem behaves, decide where tools should enter, read the market signal, then turn repeatable parts into systems.

Turn unclear problems into testable workflows

M.K. often enters AI projects in their early ambiguity: the need is not yet clear, the business model is still forming, and the team is unsure whether to start with product, content, sales or systems. The first move is not to rush into a demo, but to arrange the signals into a workflow that can be built, tested and discussed.

Find where AI should enter the process

Not every problem needs a model, and not every workflow should be automated. The work begins with data sources, decision points, delivery pressure and human bottlenecks, then determines whether AI should generate, organize, classify, send, sync replies or simply support an existing human process.

Move prototypes toward market signals

After the prototype, the real questions begin. Who replies, who tries it, who understands it, and who sees enough value to pay for it. These signals reshape the product, copy, segment and delivery model. GTM is not the last slide. It is part of product design.

Turn repeatable work into an operating system

If a project has to be rebuilt manually every time, it is difficult to turn into a business. Successful and failed steps are organized into fields, templates, naming rules, checklists, deployment flows and review rhythms, so the next cycle starts from a system rather than from scratch.

Systems

Repeatable systems move AI projects beyond one-off demos.

Recurring steps are organized into fields, templates, checklists, deployment rhythms and review methods. These systems are not meant to look polished in a deck. They are meant to be used by a team.

AI Auto BD Kits

Outbound loop / CRM / Gmail API / weekly iteration

A trackable business development system for segments, leads, sending, replies and offer iteration.

3Kup SEO delivery workflow

Website / SEO / GEO FAQ / Cloudflare / lead form

A low-friction delivery workflow for local businesses that need websites, search structure and lead capture.

Fashion Ultra production workflow

SKU intake / VTO provider / QA / cloud delivery

A controlled workflow for moving AI fashion imagery from single demos into batch production and QA.

PropertyLens report workflow

Research / buyer questions / agent trust / report templates

A research report workflow connecting real estate information, Chinese buyer context and agent trust-building.

Related Properties

Additional brands and operating surfaces.

These external sites provide adjacent context across e-commerce, content, brand systems and AI workflow experiments.

Notes

Field notes on AI adoption.

Short notes on recurring problems that appear when AI enters products, GTM and operating workflows.

Note

AI projects usually fail outside the model

The more common issues are ownerless workflows, unstructured data, unclear adoption paths, or demos that do not connect to a business rhythm.

Note

GTM starts earlier than the demo

If an AI product cannot explain who uses it first, why now, and how it will be delivered, a complete prototype may still fail to reach the market.

Note

Chinese-speaking SMB adoption needs a more grounded path

Many small businesses do not lack tools. They lack a practical bridge from existing work habits into AI-supported workflows.

Contact

For AI products, early-stage GTM, or cross-market projects between Taiwan and Canada.

The right collaboration often begins before everything is fully formed: there is a technical direction, a market signal and some internal pressure, but the work still needs to become product, workflow, content, sales and delivery systems.

EMAIL to M.K.FB Return to work