Build with AI

This guide follows the same workflow taught in the LaunchWithCursor Build with AI course. It assumes you are a non-technical Builder who wants to use AI to create websites or mobile applications — even if you are also a software engineer watching along.

By the end, you will know how to build a real product with AI coding assistants like Cursor, using a Blog Management Website as the worked example: some articles are public; others require a paid annual subscription.

Why a blog? It covers auth, content, payments, and admin flows — enough complexity to learn the workflow without drowning in edge cases on day one.

New to LaunchWithCursor? Read Our Mission first.

Five fundamentals for predictable results

Before diving into steps, here is what makes AI builds consistent instead of random:

Copy-ready prompts for planning live on Sample prompts.

Step-by-step: from idea to deployed blog

Step 1 — Install Cursor

1. Download Cursor on your laptop or desktop.
2. Subscribe on the Cursor website (the $20/month plan is enough for most Builders).
3. Open Cursor and get comfortable with File → Open Folder — you will always work inside a project folder, not a single file.

Step 2 — Create your LaunchWithCursor account

1. Sign up at launchwithcursor.com.
2. Follow Install & sign in to download the LaunchWithCursor extension (.vsix) and connect it to your account.
3. When you first open a project folder, the extension adds .launchwithcursor/ to your repo — this is how your AI assistant learns how you want the project structured and deployed, including UX guidelines and deploy rules. It also syncs key project information to the dashboard so you can understand the codebase without reading every file.

Step 3 — Write your User Story

In ChatGPT, Claude, or Gemini (not necessarily Cursor yet), describe what you want to build and ask for a User Story document in markdown.

Include:

• Product overview and goals
• User types (visitor, subscriber, admin)
• Business rules (e.g. free vs paid articles)
• User stories with acceptance criteria
• User journeys

Use the copy-ready prompt on Sample prompts — the blog example matches this course.

Save the result as USER-STORY.md.

Step 4 — Write your MVP-SPEC (PRD)

Using the same external AI tool, turn the User Story into a Product Requirements Document. On LaunchWithCursor we call this MVP-SPEC.md — it is your PRD for v1.

It should define must-haves, explicit “not in MVP” items, auth, payments, mobile strategy (PWA vs desktop-only), and success criteria.

Again, Sample prompts has a ready-made prompt.

Save as MVP-SPEC.md.

Step 5 — Create your project folder

1. In Cursor, File → Open Folder and create or select your project directory.
2. Sign in to the LaunchWithCursor extension (Install guide).
3. Confirm .launchwithcursor/pack/LAUNCHWITHCURSOR/ appeared — your AI assistant reads this for structure, skills, and deploy rules.
4. Copy USER-STORY.md and MVP-SPEC.md into the root of the project folder.
5. Commit all three to Git when ready (Git & GitHub).

On the LaunchWithCursor dashboard (after Run Full Scan), you will see your user story, spec, and — later — how database tables, API endpoints, and functions relate.

Step 6 — Create phased implementation plans

Switch to Agent mode in Cursor (so the assistant can create files). Paste the create build phases prompt from Sample prompts.

Important: the prompt tells your assistant the project will deploy on LaunchWithCursor, so plans match platform structure (monorepo, env vars, deploy checklist).

Each phase file in docs/plans/ should include:

• Goals and deliverables
• Tests & verification the AI runs before claiming done
• A ## Cursor prompt section at the bottom — copy-paste text to implement only that phase

Run LaunchWithCursor: Run Full Scan after committing plans so Spec & Plans → Build Phases updates on the dashboard.

Step 7 — Implement phase 1

Open docs/plans/phase-1.md, scroll to ## Cursor prompt, paste into a new Cursor chat, and let your assistant implement phase 1 only.

When it finishes:

• Read the implementation summary — do not skip this
• Ask questions in Ask mode if anything is unclear

Step 8 — Commit and deploy after phase 1

1. Commit and push to GitHub.
2. On LaunchWithCursor, open Deploy for your project — connect GitHub if you have not already (Deploy guide).
3. Complete Deploy → Checklist (database, env vars, domains as needed).
4. Deploy and open the live URL — see what you have so far. Early deploy catches problems while the codebase is still small.

Step 9 — Implement remaining phases

Repeat for phase 2, 3, … — one phase per chat session, using each plan’s Cursor prompt. Commit and push after meaningful progress.

Deploy again when a phase touches something visible (new pages, API, payments).

Step 10 — Phase Verify

When all build phases are done, run Phase Verify from docs/plans/phase-verify.md (copy the Cursor prompt at the bottom).

This pass re-checks:

• UX against the skills pack in .launchwithcursor/
• End-to-end flows and tests listed in your plans
• Deploy & hosting readiness

Commit, push, and redeploy on LaunchWithCursor.

You are halfway — then testing begins

If you followed solid USER-STORY, MVP-SPEC, plan phases, and verification, you have typically completed the half of the journey that used to take weeks or months and thousands of dollars before AI.

The next half is testing — making sure real people get a great experience.

Test yourself first

Click through every user journey on your staging URL: register, subscribe, read free vs paid articles, admin flows. Fix issues with your AI assistant using Copy log on Deployments or Logs when something breaks.

Hire manual software testers

Complex products have multiple user types — visitors, subscribers, admins — and you may want expert eyes beyond friends and family (who often never get past the homepage).

LaunchWithCursor trains, certifies, and equips manual software testers to exercise your product end-to-end and return structured feedback that your AI assistant can act on.

Learn how: Manual testing.

When to book a consultation

Most of the time, your AI coding assistant can unblock you. When consequences are high, or you want a human expert in the loop, book a consultation with an experienced software engineer on LaunchWithCursor.

Common reasons:

1. Validate the USER-STORY.md and MVP-SPEC.md AI helped you draft before you build
2. You are stuck mid-build and want a clear next step
3. You have an idea but do not know how to structure the product or where to start
4. You need a second opinion on architecture, regulations, pricing, or go-to-market
5. You want a path from idea → MVP without hiring a full agency
6. Deploy, DNS, or infrastructure issues need expert eyes

Join the founders community or book a session from the Book consultation page.

Quick reference checklist

Related guides

• Our Mission — why LaunchWithCursor exists
• Install & sign in — extension and starter pack
• Sample prompts — User Story, MVP-SPEC, and phase prompts
• Deploy & hosting — go live
• Git & GitHub — commits and branches
• Environment variables — secrets and config
• Manual testing — professional UX testing