# The AI Platform Blueprint

Thirteen practical artifacts for moving from an observed problem to a measured release.

Core principle: AI compounds judgment. It does not replace it.

## 01 — Idea Filter

- Name the user, painful moment and current workaround.
- Collect five direct evidence points.
- Score urgency, frequency, audience access and willingness to change.
- Reject an idea whose main proof is impressive technology.

Ship: a scored idea card with evidence, risks and a go/no-go decision.

## 02 — Problem Definition

- Name the trigger, failed outcome and consequence.
- Separate facts, assumptions and unknowns.
- Quantify the cost where evidence permits it.
- Write what would disprove the problem.

Ship: a one-page problem brief with evidence and falsifiers.

## 03 — Audience Definition

- Define role, context, skill level and constraints.
- Name the first-use trigger and the audience's own language.
- List where this audience already learns, searches and asks for help.
- State who the first version is not for.

Ship: a primary-audience card and exclusion list.

## 04 — Jobs To Be Done

Complete: “When ___, help me ___, so I can ___.”

- Name the functional progress.
- Add emotional or social dimensions only when evidenced.
- Map push, pull, habit and anxiety.
- Choose one job for the first loop.

Ship: a testable JTBD statement with switching forces.

## 05 — Product Boundary

- Write one promise, one user and one activation moment.
- Separate product, service, community and future platform ideas.
- List irreversible or trust-sensitive actions.
- Create non-goals for attractive adjacent features.

Ship: a product boundary with promise, activation and non-goals.

## 06 — MVP Cut

- Map entry, core action, useful result and next step.
- Include validation, empty, loading, failure and retry states.
- Defer anything that does not unlock the first result.
- Test with realistic content.

Ship: a vertical-slice map with included states and deferred scope.

## 07 — Architecture Brief

- Draw the browser → server → storage → provider path.
- Separate public, authenticated and administrative surfaces.
- Name each critical source of truth, fallback and retention rule.
- Record security boundaries and the simplest rollback.

Ship: a one-page system map and architecture decision record.

## 08 — AI Agent Brief

- State the outcome, architecture context and files in scope.
- List invariants, prohibited actions and acceptance checks.
- Require inspection before editing.
- Keep the change small enough to review.

Ship: a reusable task brief with constraints and verification commands.

## 09 — Definition of Done

- Define functional, content, accessibility, privacy and security criteria.
- Require type, test, build and journey evidence.
- Include mobile widths, error recovery and keyboard behavior.
- Name remaining limitations and owners.

Ship: a repository-ready Definition of Done.

## 10 — Launch Checklist

- Verify access, secrets, routes, forms, metadata, robots and sitemap.
- Test incognito access and production persistence.
- Prepare monitoring signals and a rollback decision.
- Launch to a specific audience with one primary CTA.

Ship: a signed launch checklist with evidence and rollback trigger.

## 11 — Distribution Map

- Choose one primary channel and one durable secondary channel.
- Map audience questions and search intent to content and next actions.
- Give each indexable page a distinct purpose and canonical URL.
- Add stable campaign and creative identifiers before publishing.

Ship: a four-week content, SEO and channel map.

## 12 — Feedback Loop

- Define landing, activation, lead, checkout and verified-purchase events.
- Keep personal data out of analytics payloads.
- Ask repeatable questions after meaningful moments.
- Review conversation beside funnel behavior.

Ship: a versioned event taxonomy and weekly feedback ritual.

## 13 — Iteration Scorecard

- Record the baseline and one leading indicator.
- Write a falsifiable hypothesis and threshold before changing the product.
- Change one meaningful variable for a defined window.
- Keep, kill or revise the change and record the lesson.

Ship: a weekly bottleneck, hypothesis, evidence and decision scorecard.

## Your next 60 minutes

1. Spend 20 minutes on the Problem Definition.
2. Spend 20 minutes on the MVP Cut.
3. Spend 20 minutes on the Architecture Brief.

Then decide whether the idea deserves code.

— Synapedia BUILD · https://build.synapedia.com/build