Question 1

What is VibeToSpec?

Accepted Answer

VibeToSpec turns an app idea into a blueprint an AI builder can follow: a product plan (PRD), a data model, a step-by-step build plan, acceptance criteria, edge cases, a decision log and a rulebook (AGENTS.md). It trims the idea to a first version before any code is written, checks the blueprint for contradictions, and checks your AI's work against it as you build.

Question 2

I can't code. Is this for me?

Accepted Answer

Yes, it's built for you. You describe your idea and make a few choices in plain English. The technical parts are written for your AI builder, and the Build Review tells you in plain words whether the AI's work matches your plan.

Question 3

How much does it cost?

Accepted Answer

Scanning is free, as often as you like. A complete blueprint is $19, once, and it's yours to keep. When you start building, Build guidance is $19 a month: the next instruction for every task, plus 50 result reviews and 3 workflow plans each billing period. It's bought separately and renews monthly until you cancel. Buying a blueprint never starts it, and cancelling keeps your blueprints. It doesn't include new blueprints or blueprint credits, blueprint rewrites, or unlimited AI use, and nothing is charged beyond the monthly price. 14-day money-back guarantee on the blueprint.

Question 4

What's in a blueprint?

Accepted Answer

Every file your AI builder needs: PRD.md (the product plan), the data model (for example schema.prisma), tasks.md (the build order), ACCEPTANCE.md (the definition of done), EDGE_CASES.md (the what-ifs), DECISIONS.md (your decisions, on record), AGENTS.md (the rulebook), CLAUDE.md and GitHub Copilot instructions, plus the Build Review, which checks your AI's work on GitHub.

Question 5

Do I need an account?

Accepted Answer

No. Pay with Apple Pay, Google Pay or a card, and we email you a link that downloads every file in one click. We also keep it in a free account under that email, so you can come back any time with a one-tap link: no password.

Question 6

What's an AI builder?

Accepted Answer

An app that writes software when you describe what you want in plain language: Lovable, Claude Code, Cursor, Codex, GitHub Copilot and Windsurf are popular ones. They're remarkably capable, but they can only build what they understand. Your blueprint makes sure they understand.

Question 7

What do I do with my blueprint?

Accepted Answer

Press Start building and pick your AI builder: it opens with a first message made from your blueprint. For Claude Code, Codex and Cursor, you download the blueprint and open its folder; Lovable gets the plan in its first messages. Your AI then follows the rulebook, works through the build order one step at a time, and checks each step against your definition of done.

Question 8

What is Build guidance?

Accepted Answer

Build guidance takes you through your blueprint one task at a time, with the AI builder you already use. You get the next instruction for every task. Paste what your builder replied, or the error it showed, and VibeToSpec explains what it means and gives you one next action that stays inside your approved plan. Each billing period includes 50 result reviews and 3 workflow plans; a request that fails on our side doesn't count. It keeps your decisions and progress together, walks you through checking each workflow, and writes a report. It's $19 a month, bought separately from your blueprint, and renews until you cancel. It can't see or control your builder, and it never marks something as working without your own check.

Question 9

What if my idea changes?

Accepted Answer

It will, and that's fine. For a small change, tell your AI builder and note it in DECISIONS.md. For a bigger one, scan your updated idea and create a new blueprint. Founders with a Builder Pass can still change their blueprints in place, with every version saved.

Question 10

How is this different from asking ChatGPT or Claude?

Accepted Answer

Chat assistants agree with you, so your plan stays big and fuzzy. VibeToSpec pushes back: it cuts what would slow you down, records your decisions, checks the blueprint for contradictions, and keeps checking your AI's work after you start building.

Question 11

Is this spec-driven development?

Accepted Answer

Yes, for people who don't code. Spec-driven development means writing down what to build before an AI agent builds it. VibeToSpec writes that spec for you from a plain-English idea, keeps it consistent, and makes your AI builder follow it.

Question 12

Does it check what my AI actually built?

Accepted Answer

Yes. Every blueprint includes the Build Review, which runs on GitHub each time your AI proposes a change. It runs your project's own checks, then compares the code with your plan: data your plan never approved, features you cut showing up anyway, and edits to your plan or decisions. Your AI can run the same review itself before it calls a task done.

Question 13

Which AI builders does it work with?

Accepted Answer

Start building hands your blueprint to Lovable, Claude Code, Codex or Cursor with a first message ready. Claude Code, Cursor, Codex and Windsurf read the blueprint's rulebook automatically, GitHub Copilot gets its own copy, and everything is plain text, so it works with any AI builder that can read files.

Question 14

Is my idea private?

Accepted Answer

Your idea is used only to create your results: it's sent to our AI provider to write your blueprint and is never used to train AI models. Your blueprints are saved to your account.

Question 15

What technology can my app use?

Accepted Answer

Most modern web stacks, like Next.js, React, Supabase, Firebase or FastAPI. We recommend one that fits your idea, and you can keep it or pick your own. The blueprint adapts to your choice.

Qty	Part	What it does	Saves you from
1	The product plan PRD.md	What your app does in version 1, its journeys and screens, and just as important, what it doesn't do.	features you never asked for
1	The data map schema.prisma or your stack's equivalent	Everything your app keeps, like customers, bookings or payments, and how they connect.	a database that quietly breaks
1	The build order tasks.md	A step-by-step list, so your AI builds one thing at a time instead of everything at once.	half-finished everything
1	The definition of done ACCEPTANCE.md	Plain-English checks for every feature. Nothing counts as finished until it passes them.	“done” that isn't
1	The what-ifs EDGE_CASES.md	What should happen when two customers want the same slot, a payment fails, or someone tries to see another person's data.	bugs your customers find first
1	Your decisions, on record DECISIONS.md	Every choice you made, marked as yours, and every question still open. Your AI treats your decisions as rules.	your idea being quietly rewritten
1	The rulebook AGENTS.md with CLAUDE.md and Copilot instructions	Tells any AI builder how to work on your project: what's decided, what's off-limits, and when a task is truly done.	an AI that improvises
1	The Build Review runs on GitHub	An automatic check each time your AI builder saves a change: does the work still match your blueprint and your decisions?	trusting your AI blindly

Give your AI builder a blueprint. Know what to ask it next.

Your AI builder can't read your mind, so it guesses.

What happens after you click “Scan my idea”

Describe

Understand

Decide

Blueprint

Build

What's in the box: your blueprint's parts list

You don't have to read code to know when it drifts.

Start with a blueprint. Keep building with guidance.

Free

Blueprint

Build guidance

Questions founders ask first

Bring the messy version. We'll draw it up together.