How much does it cost to finish an AI-built app?
Cost is not driven by the number of visible screens, but by what must be verified, connected and made reliable behind the interface.
The factors that determine the cost of completing an app started with Lovable, Bolt, Replit or another AI tool, from technical diagnosis to release.
Step-by-step procedure
Do not start with the number of screens. Estimate the real data flows, external services and how much existing work can be reused without carrying hidden risk. Use the steps below to turn an attractive demo into a scope that can be checked, compared and priced.
- 01
Freeze the current scope
List what must work in the first delivery and separate future ideas. An estimate keeps moving when the expected outcome is not stable.
- 02
Assess the existing project
Review installation, build, code quality, account ownership, data, security and dependencies. Mark what is reusable and what remains uncertain.
- 03
Break work into flows
Estimate sign-up, the core flow, payments, notifications, administration, migration and publishing separately, including tests and failure handling.
- 04
Define phases and acceptance
Group essential functions first and later work second. For each phase, state deliverables, exclusions, required materials and the acceptance criterion.
Never send passwords, tokens, .env files or private API keys through public forms, chats or screenshots.
Diagnosis comes first
A credible estimate needs source code, build status, connected services and a clear list of what already works. Without these, only a very cautious range is possible.
What to check
- Repository quality and accessibility.
- Number of integrations and environments.
- Current state of accounts, data and security.
Real flows contain the hidden cost
Login, payments, notifications, migrations, roles, stores and existing data require testing and failure handling. Two visually similar apps can have very different complexity.
What to check
- Count critical flows, not only pages.
- Separate fixes from new features.
- Include testing and release in the scope.
Ask for a verifiable scope
A professional proposal should define inclusions, assumptions, exclusions, required materials and acceptance criteria. Technical unknowns can be handled through a separate audit phase.
What to check
- No promises before verification.
- Split priorities into essential and later.
- Write down deliverables and responsibilities.
How to decide the next step
A useful estimate connects each cost to an observable result, separates repairs from new features and states the unknowns. When essential information is missing, the correct first step is a limited diagnosis.
Do you have an app that stops at the demo stage?
Describe what works, what is missing and the outcome you need. The request is reviewed personally by Giambattista.