Software Development, System Integration and AI Agents

01 · Life & Companionship

Meet each other as your cats

What if your cat made the friends?

Where it started: It starts with real cats and puts pet-raising play, room decorating, and socializing into one small world.
The problem: People-centered social apps can create pressure to perform. Pet photos are often shared once, with no ongoing way to interact.
Our approach: Turn cat photos into pixel characters with feeding, care, and room life. Visits, posts, and comments let interaction start from your cat's daily routine.

02 · Life & Companionship

Make looking at clouds a daily habit

What if you could keep a little piece of sky every day?

Where it started: Busy days and screens fill our lives, yet the sky keeps changing. This idea hopes to make noticing it a habit again.
The problem: Feelings in the moment fade fast, and a full diary feels like a chore. People need a way to record things without social pressure.
Our approach: Collect clouds with photos, plus the time, weather, a traced outline, and one line of text. Together they form a cloud diary you can look back on, keeping the focus on how you felt.

03 · Life & Companionship

Be together without having to chat

What if friends had a room that's always there?

Where it started: Living apart means fewer chances to just sit together casually. Company doesn't always need a formal call.
The problem: Chat rooms expect constant replies, and video calls demand attention. Neither fits the need to simply be quietly together.
Our approach: Plan a personal pixel room where close friends can drop in, move around, sit down, or leave a note, exploring everyday company through low-pressure shared presence.

04 · Life & Companionship

A place to talk slowly

What if life updates didn't need an instant reply or a public post?

Where it started: Between family, friends, and partners, some things are best said slowly, and are worth keeping.
The problem: Instant messages get buried, public social media isn't right for every private update, and writing long posts can feel like a barrier.
Our approach: Keep text, photos, and replies in separate private notebooks. Explore voice input and AI-drafted summaries that the writer confirms before sharing.

05 · Life & Companionship

Pass one diary around, turn by turn

What if a small circle of friends had a gentle rhythm of turns?

Where it started: It brings the rhythm of school-days exchange diaries back to friends now spread across different lives.
The problem: Group writing often stalls when no one picks it up. Private content also needs clear rules on who can read and whose turn it is.
Our approach: Plan private diaries, friend invites, and a fixed rotation with hand-offs, reading, and comments, so each round of exchange becomes a reason to stay in touch.

06 · Life & Companionship

What's for dinner? Check the fridge first

What if dinner's answer was already in your groceries?

Where it started: This meal tool starts from choosing today's meal, linking what's in the fridge to what you can cook.
The problem: Not knowing what you can cook, or what's missing, is a closer-to-life problem than lacking a weekly meal plan.
Our approach: Link the flow as fridge ingredients → possible dishes → today's meal → the week's plan. Soon-to-expire items and missing ingredients factor into choices, and AI suggestions are previewed before you confirm.

08 · Business & Services

Wired Office Co. (上班有線公司)

What if an online workspace let teamwork happen naturally?

Where it started: Let independent workers bring their real skills online, work side by side, and see each other.
The problem: Remote work lacks everyday time together, and tools for finding collaborators usually only get opened when a need comes up.
Our approach: Combine a shared workspace with skill showcases, talent search, and project posts, so "working together" and "finding collaborators" each stand on their own and connect naturally.

09 · Business & Services

From loving the work to booking the studio

What if booking started with understanding the work?

Where it started: An independent tattoo studio needs to show its style, understand custom requests, and schedule the right session time.
The problem: Portfolios, consultations, and bookings are scattered across different chats. Customers can't easily see the next step, and the studio has to organize everything.
Our approach: Bring portfolio display, custom consultation, session booking, and studio management into one flow, starting with a prototype to explore a single studio's real needs.

10 · Business & Services

Connect LINE chats to the ordering flow

What if ordering and back-office work connected through a familiar app?

Where it started: Businesses that use LINE as their service entry point need to connect what customers do with how orders are managed afterward.
The problem: When chat info and order status live in different places, it adds extra work to sort and search.
Our approach: Pair a LINE LIFF ordering interface with an admin dashboard and messaging events, so front-end orders and back-office handling share a clear order context.

11 · Business & Services

Let the artwork be the star of the site

What if an artist's website could stay up to date on its own?

Where it started: Artwork needs careful presentation, along with a content management approach that's easy to keep up.
The problem: If display and management are disconnected, every update may need an engineer, and content is hard to keep consistent.
Our approach: Split the work between a portfolio front end, a public content API, and an admin system, giving visual presentation and everyday updates each their own clear entry point.

12 · Learning & Creating

Make a story you can interact with

What if reading a story meant walking through it yourself?

Where it started: A narrative website built from chapters and scenes, exploring how visuals, interaction, and reading pace relate.
The problem: Static pages struggle with scene changes, and interactive design must still keep the story flowing.
Our approach: Split the entrance, chapter selection, and main experience into clear pages, using scene visuals and interactive elements to draw readers into the story step by step.

13 · Business & Services

KiWEDOGO (奇異狗之) Official Website

What if getting to know a company started with a quick game?

Where it started: The brand wants to express companionship, curiosity, and growing together, and hopes visitors can feel its playfulness right away.
The problem: Long service descriptions rarely make a brand memorable. The brand's values need an entry point people can experience firsthand.
Our approach: Open with a plane shoot-'em-up game, followed by a simple brand story and case studies, connecting "Keep we do, keep we go" to concrete ideas.

14 · Learning & Creating

Find your next song by mood

What if describing a feeling could find you music?

Where it started: Music discovery isn't only about song titles and artists. It can start from a mood, a setting, or a feeling that's hard to put into words.
The problem: Regular keyword search can't express a mix of emotion and style, and users may not know which tags to search for.
Our approach: Plan to map song traits and user descriptions onto dimensions like mood and style, then recommend songs by semantic matching, so finding music feels more like describing it naturally.

15 · Learning & Creating

Turn one phrase into a story

What if learning English started with a scene you understand?

Where it started: Short videos and images offer another way to learn a language. Phrases need context to stick.
The problem: Turning one lesson topic into copy, storyboards, narration, and subtitles leaves content scattered and hard to keep consistent.
Our approach: Build a complete content package from a single phrase, linking a scene script, storyboard, narration, and subtitles so the teaching idea flows into production.

16 · Learning & Creating

Look up a word without losing your train of thought

What if a dictionary lookup was just a tiny step in your reading?

Where it started: When reading English, you often just need to understand one word quickly and get back to what you were doing.
The problem: Switching apps and digging through pages of results breaks your reading, and a translation alone may not show how a word is used.
Our approach: Open a small lookup window with a shortcut, showing part of speech, a Mandarin definition, and example sentences in context, so you can read everything in one place.

17 · Learning & Creating

Turn a song into practice material

What if practicing with backing tracks meant less file wrangling?

Where it started: Singing practice uses the original track, the vocals, and the backing track, and all of it needs a consistent way to stay organized.
The problem: Handling audio files, naming them, and saving them in different places by hand is repetitive and makes it hard to pick practice back up.
Our approach: Starting from local audio files you're licensed to use, design a flow for source separation and practice-folder output, turning material prep into a repeatable task.

18 · Everyday Tools

Keep today's to-dos in front of you

What if reminders lived on your desktop, not buried under messages?

Where it started: Daily planning often just needs to answer: what do I do, and when, today and tomorrow?
The problem: Big planning tools are too heavy, ordinary notifications are easy to miss, and unfinished items need somewhere to stay.
Our approach: A small, always-on Mac window for scheduling. When it's time, it shows pending items with done and snooze actions, and keeps a record of your schedule and actions.

19 · Everyday Tools

Put Mandarin–English switching at your fingertips

What if switching input methods didn't mean guessing shortcuts?

Where it started: When working across desktops and remote tools, Mandarin and English input status needs clearer cues and controls.
The problem: Shortcuts don't always work well in every situation, and users may not be sure which input method is active.
Our approach: A native Mac floating icon shows Mandarin or English status, with click-to-switch and drag-to-place, exploring more direct input control.

20 · Everyday Tools

Bring the monthly plan down to today

What if your monthly, weekly, and daily plans told the same story?

Where it started: Personal work means going back and forth between long-term direction and what you can do each day.
The problem: Monthly, weekly, and daily plans sit apart, tasks get maintained more than once, and overdue work has no clear way to be rescheduled.
Our approach: Use one set of task data across monthly, weekly, and daily views, supporting breaking work down, rolling it up, and reviewing overdue items, so plans can keep adjusting.

21 · Everyday Tools

See where every purchase belongs

What if your credit card bill became a clear view of your life?

Where it started: Personal spending, shared household costs, and work expenses paid upfront can all show up on the same bill.
The problem: Knowing how much you spent isn't enough. You also need to sort out purpose, who it belongs to, and the matching receipts.
Our approach: Import credit card data to build spending records, add type suggestions, manual confirmation of who it belongs to, and receipt links, so categorizing and tracing work together.

22 · Everyday Tools

Give reconciliation a traceable path

What if categories, rules, and search all lived in one accounting workbench?

Where it started: Sorting out credit card expenses means importing, filtering, categorizing, and checking again and again.
The problem: Purchase dates and posting dates answer different questions, and mixing them makes monthly accounts hard to follow. Categorization rules also need to be reviewable.
Our approach: A transaction workbench that brings together import, rules, search, and categorization, separates the different date views, and keeps a way to check back against the source.

23 · Everyday Tools

Record each purchase only once

What if receipts, emails, and bills all pointed to the same transaction?

Where it started: Clues about spending are scattered across paper receipts, emails, credit card bills, and manual notes.
The problem: One purchase may leave several notices. Treating each as a transaction double counts, while bills alone lack item details.
Our approach: Plan around the real purchase as the core, linking multiple sources together, with data recognition, deduplication, amount checks, and monthly summaries.

24 · Everyday Tools

Make shared costs and reimbursements easier to sort

What if monthly expenses didn't mean re-identifying every line?

Where it started: One personal bill may need to be split into your own ledger, shared costs, and work expenses paid upfront.
The problem: Repeated exports, manual checkmarks, and receipt sorting make it easy to lose track of which month and purpose each item belongs to.
Our approach: A reusable flow for bill export, category confirmation, and receipt sorting that keeps human decisions and source links, then turns them into reports for each purpose.

25 · AI & Ways of Working

Turn "I want…" into a clear brief

What if a requirements doc could grow out of a conversation?

Where it started: Many products start as fuzzy ideas, and the people behind them aren't always familiar with spec documents.
The problem: Jumping straight into building can skip over goals, constraints, and use cases, and later discussions are hard to pick up.
Our approach: Use AI conversation to help clarify needs, then organize them into a structured summary that can be versioned, shared, and exported, so the idea takes shape step by step.

26 · AI & Ways of Working

Turn what was said in meetings into next steps

What if a recording left behind next steps, not just a file?

Where it started: Meetings and discussions hold requirements, decisions, and to-dos, but that info often stays stuck in audio files.
The problem: Re-listening takes time, and a transcript isn't the same as clear meeting notes. Recording, transcription, and analysis need to connect.
Our approach: A local-first workspace that links recording, transcription, transcript export, and analysis, turning meeting content into material you can review and organize.

27 · AI & Ways of Working

Back-office support for a one-person studio

What if designers could save their energy for direction and creative work?

Where it started: Freelance design work involves more than design: requirements, planning, quotes, purchasing, follow-ups, and handoffs.
The problem: When one person switches between many roles, project info and follow-up work end up scattered across documents.
Our approach: Organize work by project stage and department role, with AI helping on documents, tracking, and cross-project summaries, while the founder keeps direction, design, and client sign-off.

28 · AI & Ways of Working

Give AI-assisted development a process that carries on

What if every build left a starting point for the next one?

Where it started: AI-assisted development isn't just generating code. It also covers understanding requirements, design, review, and testing.
The problem: A single conversation easily loses context, and work across different roles needs a shared delivery structure.
Our approach: Explore a staged process from requirements to verification, turning roles, specs, reviews, and knowledge into reusable methods that build up experience across projects.

29 · AI & Ways of Working

Turn AI collaboration into trackable work

What if long tasks always knew where they stood and what's next?

Where it started: Software that keeps evolving needs tasks, tools, and verification data to share a common context.
The problem: When work is interrupted, versions change, or tasks are handed off, conversation memory alone can't confirm what's truly done.
Our approach: Link delivery through work items, skill-based roles, execution logs, and verification gates, keeping recoverable progress and checked knowledge.

30 · AI & Ways of Working

Give AI actions clear boundaries

What if every change came with a known scope and reason?

Where it started: Once AI can work with files and tools, the scope of work, responsibility for changes, and acceptance criteria need to be spelled out.
The problem: When sources change or several parties edit at once, mismatched states are easily mistaken for the same result.
Our approach: Use a clear work scope, a single owner for changes, independent review, and evidence checks to explore development collaboration that balances speed and traceability.

31 · AI & Ways of Working

Let automations reach a person when they're done

What if small tools didn't each have to rebuild notifications?

Where it started: Many kinds of automation need to send a notice when they finish, hit an error, or need help.
The problem: When each project connects to messaging services on its own, recipients, content, and usage get handled over and over, and are hard to manage in one place.
Our approach: Turn message sending into a shared entry point, starting with LINE notifications, with previews, chosen recipients, and usage lookups, so tools share one consistent way to notify.

32 · AI & Ways of Working

Give every idea its own test space

What if sharing a prototype didn't mean rebuilding the setup each time?

Where it started: Multiple projects and versions each need their own testing, plus an easy way to share previews.
The problem: When services, ports, and data are mixed together, they can interfere with each other, and updates need to know which source they came from.
Our approach: Organize previews with separate test spaces, clear sources, and deployment contracts, bringing isolation, updates, and rollback into one way of working.

33 · Business & Services

Back list research with verifiable evidence

What if finding partner brands meant more than collecting names?

Where it started: Business development first needs to identify brands that meet size and channel criteria, then decide whether they're worth a closer look.
The problem: Store counts, social media reach, and brand identity are scattered across outside sources, and the data may be outdated, missing, or conflicting.
Our approach: Design a criteria-based brand discovery flow that gathers public info, keeps observation dates and sources, and sends anything unconfirmed to human review.

34 · AI & Ways of Working

Make login and sharing a shared starting point

What if new tools didn't have to build accounts from scratch?

Where it started: Many apps need sign-up, verification, login, and a way to share content with specific people.
The problem: These seemingly repetitive features involve different permissions and invite states. Rebuilding them each time leads to inconsistencies.
Our approach: Pull email verification, login, sharing, and role permissions into standalone demo components that future products can reference and adapt to their own needs.

35 · Web3 Experiments

Connect community participation to digital rewards

What if joining an event could link to verification and digital collectibles?

Where it started: Community events by brands or creators want to link participation, eligibility checks, and the reward process.
The problem: Community tasks, event status, and on-chain actions are scattered, so users easily get lost between steps.
Our approach: Explore combining community task verification, event eligibility, and reward distribution, with wallet and digital collectible interfaces, to build one consistent participation journey.

36 · Web3 Experiments

Explore ways to share creator income

What if the rules for splitting digital work income were easier to see?

Where it started: Between trading digital works and community participation, there's room to explore different revenue-sharing models.
The problem: How revenue is collected, held, calculated, and claimed needs a clear division of roles and rules that can be reviewed.
Our approach: Use modular contracts to explore royalty collection, asset custody, and distribution, breaking the rules into clear responsibilities. This only describes the design concept.

37 · Web3 Experiments

Another form for software licenses

What if users could hold and transfer their own licenses?

Where it started: Software licenses are usually tied to platform accounts. This idea explores a different way to hold them.
The problem: Developers keep rebuilding account and license systems, and users' rights to use services lack a shared form across services.
Our approach: Plan for digital credentials that represent the right to use a service, with wallet verification and a shared license interface, exploring how holding, verifying, and transferring relate.