Skip to main content
Product galaxy

AI Developer
Program

Take your Vibe Coding live

From specs and AI-assisted development to data, permissions, testing, debugging, deployment and rollback, put your work in the hands of real users.

NT$ 4400 NT$ 9980
See plans
Starts2026/11/30
ForFor people who have built a Demo with AI but haven't launched it yet
Courses8 courses
TeachersHai Tseng、Eddie Kao、Eric Wu、Allen Own、Richard Lee
POINT 01

Is your next AI-built product
ready for other people?

Three stages of the same app: a sketch on paper, a Demo that runs only on your computer, and a product a second person can open too localhost:3000 · Just you clients.myapp.tw

From spec to deployment, follow the teachers and ship a product to a URL others can open

POINT 02

Vibe Coding gets you to 80%.
Why does it stall there?

Sound familiar?

AI fixes one thing and breaks another. When does it end?

In this course:Lessons 03, 06 | Save versions with Git to roll back, and catch breakage with tests

It works fine on my computer, but my friend can't open it or log in.

In this course:Lesson 07 | Deploy to a URL others can open

If AI can query members and orders, will it see fields it shouldn't?

In this course:Lesson 04 | Limit what AI can query and verify the results

The API key sits right in the code, and you'd never know if it leaked.

In this course:Lesson 05 | Manage Secrets and run a leak check

After connecting payments, have you tested success, failure and duplicate notifications?

In this course:Lesson 07 | Run pre-launch cases through test payments

AI says it's done, but give it the same prompt again and you get something different.

In this course:Lesson 02 | Write the spec first and define what done means

The last 20% is specs, data, permissions, testing and deployment.
You can learn every one of them

POINT 03

AI writes the code.
You lead AI to turn it into a product

01SpecSDD
02Build in stagesTask list
03TestingCheck quality
04VersionsRestore with Git
05DeployA second person tries it

You lead AI
and take a product from spec to launch.

POINT 04

What stands between
your Demo and a real product?

A maker at a laptop on the left with a finished Demo, a closed door in the middle, and outside it a person holding a phone, waiting to try it STARTWorking Demo GOALSecond user
01Product specScopeRequirementsUse cases
02Data safetySchemaQuery accessResult checks
03Secrets & permissionsSecret storageLeast privilegeLeak check
04Version controlChange historyBranchesRollback
05Testing & debuggingHappy pathError casesRegression tests
06Deploy & launchPublic URLProductionLog in
GOALA second user can try it smoothlyUnderstandUseFinish the core task
POINT 04

Know what each gate needs and how to check it, and you'll see what your product still lacks

POINT 05

How far can one person
take a product?

In class, the teachers built these with the same method:

A maker standing by a desk holding up a checklist, with a large screen showing the products built in class

Watch the teachers do it step by step, and you'll know where to start your own product

POINT 06

From idea to launch,
six skills to lead it solo

A maker in a spaceship cockpit, a flagged planet outside the window, and six screens around: a spec sheet, a handoff between two AIs, a database with a magnifier, a shield with lock and key, branches with a rollback arrow, and a cloud with servers and a globe

You understand the purpose, risks and definition of done at every step, and know how to have AI handle it

POINT 07

Six stages
to launch your first product

  1. 01KickoffA spec and task list AI can readLesson 2
  2. 02BuildA working prototypeLessons 2, 3
  3. 03SetupAccurate data, well-kept keysLessons 4, 5
  4. 04TestingPassing tests + versions you can restoreLessons 3, 5, 6
  5. 05LaunchA shareable, working URLLesson 7
  6. 06TrialA second person uses it + recovery after a breakLesson 8
Lesson 08 One small product goes from stage 1 to stage 6
Follow along with your own: work tools · interactive content · service products See all eight lessons →
POINT 08

No engineering background?
Can I really do this?

A maker stepping out of a laptop, handing the finished app to a second person

Solid: lessons available now Dashed: lessons coming soon

Lesson 01 | HaiNot one step in the process requires codingIn Lesson 01, Hai says no step in this process requires you to know how to code
Lesson 02 | Eddie KaoA 45-line prompt builds an expense appHe leads AI from an empty folder to a finished app, and says anyone can write a spec
Lesson 08Break it, and you can still recoverThe lesson demos everything from a blank project to launch, including bugs, breakage, rollbacks and deployment snags
Lesson 08A second person can open itThe lesson demos full proof of launch: a URL that opens, working features, test records and a second person's trial

Walk through a real development process once with a guide, and your first product can leave your computer.

POINT 09

Every launch checkpoint
has a teacher in charge

  1. Hai TsengHai TsengLessons 01, 08The big picture
  2. Eddie KaoEddie KaoLessons 02, 03SpecVersion control
  3. Richard LeeRichard LeeLesson 06Testing & verification
  4. Allen OwnAllen OwnLesson 05Security
  5. Eric WuEric WuLessons 04, 07DatabaseDeploy

Five teachers each own one checkpoint. Finally, one small product runs through all six stages

POINT 10

You can already build your ideas.
Now finish one for real.

Vibe Coding lets anyone start building. This course gives you the skills to actually finish

An engineering blueprint spread in front of an open course box, linking spec, versions, data permissions and secrets, testing, deployment and a second user, with an expense app, a booking site homepage and a query tool in front of the box
  • 8lessons
  • 6stages to follow
  • 1shareable URL
Launch price NT$ 4400 NT$ 9980

Bringing your team?Learn about team plans →

COURSE8 lessons · 5 teachers
  1. 01

    Software development in the AI era: From spec to deployment

    See the full path from spec through development and testing to launch, and spot what your project still needs.

    Hai Tseng
  2. 02

    Lead AI with a spec: Spec-driven development (SDD)

    Write the spec, split the tasks, let AI build to it, then check each item.

    Eddie Kao
  3. 03

    Version control with Git: Track changes, isolate risk, roll back

    Save versions and branch with Git, so you can recover when something breaks.

    Eddie Kao
  4. 04

    Let AI query databases safely: Permissions, field meaning and result checks

    Limit what AI can query, understand what each field means, and check the results.

    Eric Wu
  5. 05

    Security in practice: Secrets, permissions and launch safety

    Protect keys and permissions and run a security check before launch.

    Allen Own
  6. 06

    Software testing with AI: Tests, code review and a regression harness

    Write tests for core features and rerun them after every change, so old features keep working as you add new ones.

    Richard Lee
  7. 07

    Going live: Deployment, login and payments

    Deploy to a URL others can open, with login and test payments.

    Eric Wu
  8. 08

    From zero to one: Ship your first product

    Take a small product from spec to launch with the instructor, and have a second person use it.

    Hai Tseng
Lesson 08 full demo

Get a second user
to finish one task

The lesson starts with a small web product and works through spec, development, data, security, testing and deployment. Then someone who took no part in building it opens the URL and completes the core task. Use these nine checks on your own product.

01Spec 02Core feature 03Data 04Security 05Versions 06Testing 07Deploy 08Seconduser 09Rollback 0/9

A second user completes the assigned core task

  1. 01A clear MVP spec and user tasks
  2. 02At least one core feature that does real work
  3. 03Safe test data you can query and verify
  4. 04Security checks on permissions, secrets and sensitive data
  5. 05Version control, with versions you can roll back to
  6. 06Tests for core features and failure cases
  7. 07Deployed where other people can use it
  8. 08A second user completes the assigned core task
  9. 09Break it on purpose, then return to a working version
Q&A
  1. 01

    Who is this for?

    People who have tried Vibe Coding, maybe even built a Demo, but haven't launched it yet. For example: PMs and designers who want to build a testable product themselves, founders and solo businesses validating ideas on a small budget, people turning forms and processes into internal tools, and teachers and creators turning their expertise into interactive lessons or quizzes.

  2. 02

    Do I need other galaxies before Workflow?

    No.Each galaxy is a standalone series. Pick by the work you want to get done.

  3. 03

    Do I need to know how to code?

    You don't need to become an engineer first.The course covers the software concepts you need, from specs, versions and data to permissions, testing and deployment, so you can judge what AI hands back.

  4. 04

    How is this different from Maker, Twin and Workflow?

    You build a product other people can use.Maker helps you create high-quality cross-media work. Twin lets an AI assistant take over your personal tasks. Workflow makes an AI workflow deliver reliably. Product takes a small web product to the point where real users can try it.

  5. 05

    How big a product can I build?

    A small web product.One with clear core features that can be tested and opened to the public or a controlled group of testers. Large-scale system architecture and enterprise operations are outside the scope of this course.

  6. 06

    Can I take the course on my phone? What computer do I need?

    A computer is required.Phones and tablets work for watching but not for following along. Your computer must allow you to install software, so check installation permissions on a work computer first. Claude Code needs macOS 13 or Windows 10 (version 1809) or later and at least 4GB of RAM; the Codex desktop app needs an Apple silicon (M-series) Mac, or on Windows, install it from the Microsoft Store; Spectra supports Apple silicon (M-series) Macs and Windows, not yet Linux; update Git to a recent version. To watch lessons, use Chrome 111, Safari 16.4 (iOS 16.4), Firefox 128 or later.

  7. 07

    Which AI tools will I use? Can I switch?

    Pick Claude Code or Codex.You're not locked into one tool. You'll also use VS Code, a recent version of Git and the instructor's Simple SDD prompts; Spectra is optional. The database shown in lesson 04 is SAT. Knowledge's internal system, which students can't log in to; the instructor explains how to bring the approach into your own environment.

  8. 08

    Do I need to pay for anything else?

    Developer tools are paid.Claude Code isn't available on the free plan; it requires Claude Pro or higher, or API billing. Codex's free trial allowance is limited, so we suggest Plus or higher to follow the whole series. VS Code, Git and Spectra are free. Hosting, domains and payment services for launching may cost extra, and many offer a free tier to try. Whenever a lesson uses a paid feature, we'll flag it.

Want your whole team to join?

Learn about team plans →

Plans

This price ends in 0d06:06:10
NT$4,400List priceNT$9,980
  • AI Developer Program, 8 lessons
  • 12-month access, counted per lesson
  • Includes 1 AI skills assessment
  • 0% installments: 3, 6 or 12 months (non-local cards excluded)

Multi-buy2 items 10% OFF · 3+ items 15% OFFSee multi-buy

Courses open 2026/11/30 17:00

This price ends in 0d06:06:10

All six galaxies, one price

  • All 6 galaxies, all 35 lessons unlocked
  • 12 months of access, counted per lesson
  • Includes 6 AI skills assessments
  • 0% installments over 3, 6 or 12 months (not valid for foreign cards)
NT$12,800Bought separately List priceNT$50,090 Save NT$9,000 vs. buying separately
NT$4,400List priceNT$9,980 See plans