AI Blueprint Product Guide

AI Blueprint turns an application idea into a connected planning package that can be reviewed, revised, versioned, and handed to a developer or AI coding agent. This guide brings together the product concepts that are easiest to confuse.

1. What AI Blueprint Produces

A Blueprint is not a single chat answer. It is a package of documents built from one approved product context. The current set can include Idea Review, Application Journey, Technical Foundation, Database Schema, App Structure or App Map, Access & Security, Lean Canvas, MVP Scope, Requirement Coverage, and UI/UX Design.

Each document answers a different question:

Not Required is a valid outcome when an artifact does not apply. Open and review the current stub; do not treat it as a generation failure.

2. Lifecycle and Two Approvals

Idea
→ enrichment and critical questions
→ Idea Review / VPC Sign Approve
→ connected document generation
→ review and revise current versions
→ Sign Approve Blueprint
→ Export Project ZIP

Idea Review / VPC Sign Approve approves product context and starts generation. Sign Approve Blueprint approves the complete current version set after review and unlocks eligible export.

Reviewed only records that the current document was opened for human inspection. It is not either approval and does not prove that every review note is resolved.

3. Reading Blueprint Summary

• Current Status explains what the project needs now.
• Blueprint Documents shows readiness and review state.
• Review Progress counts current documents opened for review.
• Next Action points to the next useful customer action.
• Project Health separates notes, before-go-live work, repair states, and blockers.

A review note is not automatically a blocker. Read its owning document and classify it as a question, assumption, risk, before-go-live task, or true stop condition.

4. Revisions with Blueprint Assistant

Assistant chat can answer a question without changing a document. For a change, request a proposal and review its impact first.

1. Create a proposal when the intended outcome is known but nothing should change yet.
2. Direct edit for a precise value, label, typo, duration, or color in one leaf document.
3. Rewrite one document when one document needs new reasoning or wording.
4. Major revision and regeneration when authoritative actors, features, rules, integrations, or scope change.
5. Regeneration from a proposal only after its target and impact are clear.

Before Apply, verify the summary, scope, and affected documents. Never enter passwords, API keys, tokens, OTPs, payment credentials, or unnecessary customer data in chat.

5. Versions, Restore, and the Baseline

Revision history tracks changes to one document. Blueprint versions store whole-Blueprint snapshots. You are here marks the active snapshot.

Restore moves the whole Blueprint to an older snapshot without deleting history. Review the new current set and approve it again when export requires a fresh approval. Never mix files from two snapshots into one developer baseline.

6. Project ZIP and Coding Agents

Project ZIP contains specifications and workflow, not a finished application. It can include a consolidated PRD, Human PRD, Complete Database Schema, individual documents, UI mockup direction summary, manifest.json, safe .env.local placeholders, .gitignore, agent rules, an implementation guide, and five staged prompts.

Safe sequence:

1. extract the archive into a normal folder;
2. read manifest.json, AGENTS.md, the implementation guide, PRD, schema, and relevant documents;
3. preserve existing repository rules and source code;
4. never put real secrets in the ZIP template or Git;
5. create a Git baseline;
6. run PROMPT_1.md and review the plan;
7. advance through a bounded prototype and one production task per run; and
8. verify tests, UI, diffs, and acceptance criteria before committing.

Read From AI Blueprint Project ZIP to Code for the complete workflow.

7. Free Access, Credits, and Billing

The ordinary free preview and a campaign trial-credit offer are different mechanisms. The ordinary preview is limited and read-only. A campaign can have its own duration, credit amount, and eligibility; use what your account shows instead of relying on an older example.

AI credits are usage units, not money. Payment history records money transactions; the credit ledger records usage and restoration. Credit restoration is not a bank refund.

The current catalogue uses:

• IDR → Midtrans; Starter Rp50,000 and Plus Rp100,000.
• USD → PayPal; Starter US$8 and Plus US$15.

Add-ons follow the currency selected in Pricing or Billing. The server chooses catalogue amounts; a browser must not provide a free-form total. Entitlements activate only after backend verification of the provider's official status, not merely after a browser redirect.

8. Language, Security, and Privacy

The dominant idea language controls document output. The ID/EN toggle changes interface copy; it does not rewrite an existing Blueprint. The PRD language selects the implementation-kit language.

Use only necessary data. Remove secrets and sensitive identities from prompts, attachments, logs, screenshots, and support tickets. Health, biometric, face, voice, child, payment, and third-party identity data require clear authority, consent, purpose, access, retention, and deletion handling.

AI output requires human review. Build-ready means better structured for implementation handoff; it is not a guarantee of correctness, legal approval, production security, or a completed application.

9. Available and Coming Soon

The planning workspace, documents, review, Assistant proposals, supported revisions, versioning, downloads, and approved Project ZIP handoff form the current planning experience. Development Planning, Project Management, Deployment Assist, and the Pro plan remain Coming Soon until the product announces availability.

A Blueprint can describe payments, appointments, marketplaces, or other features for your future application. That does not make them built-in AI Blueprint modules.

10. Handoff Checklist

• Outcomes, users, scope, and non-goals are consistent.
• Requirements have testable acceptance criteria.
• Data, access, audit, failures, and recovery were reviewed.
• Every current document, including Not Required stubs, was reviewed.
• Review notes and blockers were classified.
• The active version set is correct and has Blueprint approval.
• Project ZIP comes from the intended baseline.
• No secrets appear in the ZIP, prompts, screenshots, or Git.
• The coding agent creates a plan before full implementation.

If account, payment, or export state looks wrong, keep a safe transaction or project reference and contact Support. Do not pay again or regenerate the whole project before checking the actual state.