Web & Software Development / Field guide
What It Really Costs to Build a Web App with AI (2026)
MarkPublished · Updated
AI has made writing code cheaper, not the whole job. What drives the cost of a web app, what AI really saves, what you pay every month after launch, a worked estimate for a booking app, and how to compare quotes.
The short answer: AI cuts coding costs, not decision costs
AI tools can generate a working app screen in minutes, so it's natural to expect a whole web app to cost almost nothing. Then a quote arrives and the number is much higher.
The gap comes from what a web app's cost actually is: mostly people's time, spread across six kinds of work. AI shortens some of them a lot and others barely at all.
- DecidingFeatures, user flows, what to leave outAbout the same
- DesigningScreens people can actually useA bit faster
- Writing codeScreens, logic, databaseMuch faster with AI
- ConnectingPayments, email, calendars, CRMAbout the same
- Securing & testingAccess control, data protection, test casesA bit faster
- Launching & maintainingHosting, updates, fixes, monitoringAbout the same
AI mostly speeds up writing code. The parts that decide whether an app works for the business still need experienced people.
That's why a well-built app costs less than it used to, but not "almost nothing". The parts AI doesn't speed up are the ones that decide whether the app works for your business.
Three ways to build
Your biggest cost choice is who builds it:
- You, with AI app builders: the lowest upfront cost (subscriptions plus your time), but you're the builder, tester and support. Best for prototypes, internal tools and testing an idea.
- A freelancer: a medium cost for a clear, small scope. The main risk is one person's availability and skill set.
- An agency: the highest upfront cost, the least of your time. Worth it for apps with payments, personal data or several integrations.
The subscription is rarely the real cost of the first route; app builders like Lovable list their plans on their pricing page. The real cost is your time and fixing what the tool got wrong before real users and data arrive. Our guide to building a website with AI compares the three routes in detail.
What drives the price of a web app
Whoever builds it, the same six things decide most of the price. Each one turns a small job into a big one:
AI features deserve a special mention. If your app uses a language model, for example to answer questions or summarise documents, you pay for usage every month, and the app needs protection against misuse such as prompt injection.
What AI makes cheaper, and what it doesn't
- Standard screens and forms
- Boilerplate code and setup
- First-draft copy
- Prototypes to test an idea
- Design exploration
- Writing tests
- Debugging common errors
- Documentation
- Deciding what to build
- Security review
- Integrations with real business systems
- Accountability when something breaks
This is why a weekend prototype is a great way to test an idea, and why taking it to production often costs more than founders expect.
The pattern: AI is excellent at producing standard things quickly. It's much less help with things that are specific to your business, your customers' data or your existing systems, and those are usually where the money goes.
Running costs after launch
Plan for monthly costs from day one. They're small at first and grow with users.
| Item | Examples | Grows with | Free to start |
|---|---|---|---|
| Hosting | Vercel, Netlify | Traffic | Usually |
| Database & sign-in | Supabase, Firebase | Users and stored data | Usually |
| AI model usage | OpenAI, Anthropic, Google | Text processed | No |
| Email & SMS | Resend, Twilio | Messages sent | Usually |
| Monitoring & backups | Sentry, provider backups | Events and data size | Usually |
| Maintenance | Developer hours | Age and size of the app | No |
Example providers only, not recommendations. Free tiers make an early app cheap to run; check what the next tier costs before you grow.
Hosting platforms like Vercel and database services like Supabase publish their tiers, and many start free. That makes an early app cheap to run. Read what triggers the next tier before you grow, and budget developer time for maintenance every month: security updates, dependency updates and small fixes.
The hidden costs founders miss
The most expensive costs usually appear after the first version works:
- Security fixes. AI-generated code can include common weaknesses like broken access control or exposed secret keys. The OWASP Top 10 lists the most common web app risks and makes a good review checklist.
- Rework. An app built quickly without a clear structure can be cheaper to rebuild than to extend.
- Lock-in. Some builders make it hard to export your code or move hosting. Make sure you'll own the code and the accounts.
- Your own time. Every hour spent testing and fixing is an hour not spent on customers.
A worked example: estimating a simple booking app
Here's how an estimate breaks down. Imagine a small clinic wants an app where patients book appointments, get reminders and pay a deposit, while staff manage the schedule.
- Discovery & feature listUser flows, scope, what’s outSmall
- DesignPatient and staff screensMedium
- Booking & schedule logicAvailability, cancellationsLarge
- PaymentsDeposits and refundsMedium
- RemindersEmail or SMSSmall
- Staff adminRoles, permissions, schedulesMedium
- Security & testingAccess control, data protectionMedium
- LaunchHosting, domain, analytics, handoverSmall
- Effort
- How much AI helps
Illustrative effort, not a quote. Patients book, get reminders and pay a deposit; staff manage the schedule. Most of the remaining cost sits in payments, permissions and security.
AI speeds up the screens, the first draft of the booking logic and the tests. Payments, permissions and security still need careful human work, and that's where most of the remaining cost sits.
How to get an estimate you can trust
The better your brief, the more accurate the estimate. Before asking anyone for a price, write a one-page brief. Here's one filled in for the clinic app:
- 1. Who uses it, and what do they need to do?
- Patients book and pay a deposit. Staff see and change the schedule. One admin manages staff accounts.
- 2. Must-haves for launch (and what can wait)
- Booking, deposit, email reminder, staff schedule. Later: SMS, reviews, loyalty points.
- 3. Integrations
- Card payments; Google Calendar for staff; leads into our existing CRM.
- 4. What data, and is any of it sensitive?
- Names, phone, email, appointment type. No medical records stored in the app.
- 5. Budget range and deadline
- A range, not a single number, and a launch date with the reason for it.
Send the same brief to every provider, so the quotes are comparable.
Then compare what each quote includes, not just the total:
| What’s included | Low quote | Complete quote |
|---|---|---|
| Building the agreed features | ✓ | ✓ |
| Design, not just a template | ✕ | ✓ |
| Testing, with test cases you can see | ✕ | ✓ |
| Security review before launch | ✕ | ✓ |
| Launch: hosting, domain, analytics | ✓ | ✓ |
| You own the code and the accounts | ✕ | ✓ |
| Maintenance terms after launch | ✕ | ✓ |
| Which parts AI does, and who checks them | ✕ | ✓ |
Illustrative. A lower quote often isn’t cheaper; it just leaves things out that you’ll pay for later.
The app is only part of your launch budget. Our guides on how much a startup should spend on marketing and building a marketing plan cover the rest.
Our take: pay for the parts AI can't do
AI coding tools have cut the time it takes to write standard code. The right way to use that saving isn't to skip review. It's to spend more time on the parts that decide whether an app succeeds: the feature list, the integrations, security and the launch.
When you compare providers, ask each the same question: "Which parts of this project does AI do, and who checks them?" A good answer names the tools, the human review steps and who is responsible when something breaks.
Key takeaways
- AI cuts the cost of writing code, not of deciding, integrating and securing.
- Features, user roles, integrations, data sensitivity and design drive most of the price.
- Budget for monthly running costs and maintenance from day one.
- Watch for hidden costs: security fixes, rework and lock-in.
- Send every provider the same written brief, and compare what each quote includes.