Andrew Crossley

FREE · 20 QUESTIONS · NO EMAIL

Is your AI-built app ready to launch?

A working Lovable, Bolt, Cursor or AI-assisted app is not automatically production ready. Score the product across customer evidence, access and data, reliability, AI quality and cost, and commercial launch readiness. You will get the result immediately. This is a product diagnostic, not a security certification.

PROGRESS

0 / 20 answered

CUSTOMER EVIDENCE

1. Can a target user complete the core journey without you explaining the interface?

A founder-guided demo is not the same as a product someone can use alone.

2. Have real target users outside your immediate team or friends used the product?

Matched users expose different behaviour from people who already understand what you are building.

3. Have you defined and instrumented the one event that means the product delivered value?

Examples include a completed booking, generated outcome accepted by the user, or a paid task completed.

4. Do you know whether users return or repeat the valuable task at the natural frequency of the problem?

A first use proves curiosity. Repeat behaviour is stronger evidence of lasting value.

ACCESS & DATA

1. Have authentication and server-side authorisation rules been tested deliberately?

Signing in is not enough. The system also needs to enforce what each account is allowed to read or change.

2. Have you tested that one customer cannot access another customer's private data?

Try the failure case rather than assuming generated row-level or API rules are correct.

3. Are API keys, service credentials and privileged actions kept out of client-side code?

Anything shipped to the browser should be treated as visible to the user.

4. Do you understand backup, deletion and recovery for customer data?

You should know how data is restored, removed and handled when an account closes or something goes wrong.

RELIABILITY

1. Do important failure states show a useful recovery path instead of a blank screen or raw error?

Test bad input, unavailable services, timeouts and empty states as well as the happy path.

2. Do you have structured error tracking or logs that let you reproduce a customer problem?

A screenshot from a user should not be the only evidence that something failed.

3. Will you know when the core journey is failing before several customers report it?

Use monitoring or alerts for the failures that would stop customers receiving value.

4. Is there a clear support and escalation path when the app cannot resolve a user's problem?

Early products need a human recovery route even when the long-term goal is more automation.

AI QUALITY & COST

1. If the product uses AI, do you have a representative evaluation set made from real inputs?

If AI is not part of the product, choose Yes. Otherwise use real examples rather than demo prompts.

2. Can you state what acceptable AI output looks like and measure changes against it?

If AI is not part of the product, choose Yes. Quality should be testable rather than a feeling from a few prompts.

3. Does the AI feature have a safe fallback, correction or human path when it is wrong?

If AI is not part of the product, choose Yes. The interface should acknowledge that model output can fail.

4. Do you know the AI and infrastructure cost per successful customer task or active user?

If AI is not part of the product, use the equivalent infrastructure cost. Price needs to survive real usage, retries and heavy users.

COMMERCIAL LAUNCH

1. Can a new user understand the value and reach the core journey without a founder-led onboarding call?

High-touch onboarding can be a deliberate model, but it should be a choice rather than a workaround for unclear product flow.

2. Are acquisition, activation and the main conversion events visible in analytics?

Launch should tell you where people stop, not only how many people visited.

3. Is there a real commercial test such as payment, paid pilot, qualified waitlist or another commitment signal?

The right signal depends on the business model, but the MVP should test more than whether people say the idea sounds good.

4. Have you decided which evidence will make you invest more, change direction or stop?

Write the decision threshold before launch so the result is not reinterpreted emotionally afterwards.

YOUR SCORE

Answer all 20 questions to see the result

No signup or email gate.

What this score can and cannot tell you

The score is designed to expose the gaps that are easy to miss when an AI-built prototype looks finished in the browser: customer evidence, data boundaries, failure recovery, observability, AI evaluation, cost and commercial measurement.

It is not a penetration test, legal review, compliance assessment or guarantee that a system is secure. Products handling health, financial, children’s, identity or other high-consequence data can require specialist technical, legal or regulatory review beyond this diagnostic.

If you have already built the product in Lovable, Bolt, Cursor, Replit or a similar stack, the next step should normally be to identify the specific blockers before accepting advice to rebuild the application from scratch.

EXPLORE NEXT

Continue the journey

Every stage from first idea to first revenue, mapped end to end.

Lovable App Review & Production

Audit a Lovable or AI-built app, fix the real launch blockers and take it to a production-ready release.

AI MVP Production Readiness Checklist

Seven gates that take an AI-built MVP from demo to something you can charge for.

From Prompt to Production

The five stages between an AI prototype and a product customers pay for.

AI Product Consultant

Turning LLMs, RAG and agents into products that hold up in production.