There are five realistic ways to get an app built: hire an agency, hire a freelancer, use a no-code builder, use an AI builder, or build it yourself. They differ in price by a factor of a hundred, in time by a factor of ten, and, more importantly than either, in what you are left owning when the work is done.
Most advice on this question is written by one of the five routes, which is why it recommends that route. What follows is the version you would get from someone who has been on the receiving end of all of them and then had to keep the result running.
Table of contents
- Before you talk to anyone: write the one-page brief
- Route one: an agency
- Route two: a freelancer
- Route three: no-code and AI builders
- Route four: build it yourself
- The cost that outlives the build: running it
- How this fits the rest of the stack
- FAQ
Before you talk to anyone: write the one-page brief
Every route gets cheaper and faster if you do this first, and every route gets expensive if you do not. One page. It contains:
- The types of user, and what each one needs to be able to do. Two or three types is normal. Eight means you have not decided what the app is.
- The screens, drawn in pencil. Boxes and arrows are enough. The drawing is not the design, it is the proof that you have thought about the flow.
- The one thing it must do on day one. Not the roadmap. The single action that, if it works, makes the app worth having.
- What it connects to. Payments, email, a spreadsheet you already have, a calendar, a supplier’s system. Integrations are where quotes double.
Without this page, every quote you receive is a guess dressed as a number, and the guessing is done by the party who benefits from guessing high.
Route one: an agency
An agency gives you a team: design, development, project management, testing. For a real product with real users on day one, this is the route that most reliably ships something finished. It is also the most expensive by a wide margin, and the timeline is measured in months.
The risk is not the bill. It is ownership. Agencies routinely set up the hosting, the domain, the database and the app store accounts under their own credentials, because it is faster for them. Six months later you have an app you cannot deploy, on infrastructure you cannot log into, written in a stack nobody on your side can read.
Three questions to ask before signing anything: where will this run, whose name is on every account, and can I deploy a change myself on the last day of the contract. If any answer makes them uncomfortable, the contract needs a clause.
Route two: a freelancer
A good freelancer costs a fraction of an agency and moves faster, because there is no project manager between you and the person typing. For a first version with a clear brief, this is often the best value on the list.
The trade is that you have one person. If they are unavailable, the project is unavailable. So the rules are the same as with the agency, applied harder: the code lives in a repository in your account from the first commit, the app is deployed on infrastructure you control, and there is a written note of how to build and deploy it that someone else could follow.
Pay for the handover explicitly. An hour of a freelancer walking a second developer through the project is the cheapest insurance you will ever buy.
Route three: no-code and AI builders
No-code builders let you assemble screens and data from a browser. AI builders go a step further and generate the application from a description. Both produce a working draft in an afternoon, and both are genuinely useful for finding out whether the idea is any good before spending real money on it.
Then the app asks the questions the draft skipped. Where do the form submissions go. Who can log in. Where are the uploaded files. What happens when someone clicks twice. Your AI-generated form was cute until it asked where the submissions go, and the answer to that question is a backend, a database and somewhere for both to run.
The question to ask any builder is whether you can take the result out. Some no-code tools hold the app inside their platform permanently, which is fine until the pricing changes. AI builders usually give you code, and code is portable, but it has to be deployed somewhere, and that somewhere is a decision you are making whether you notice or not.
Route four: build it yourself
This is more achievable than it sounds, for one reason: most apps do not need to be native apps. A web application that works well on a phone, can be added to the home screen and sends notifications covers the majority of small business ideas, and a web application is a much smaller thing to build than two native apps plus a backend.
A web app, structurally, is one deployable service, one database, some file storage and a domain. That is a shape one person can build, deploy and understand. The learning curve is real, but it is a curve rather than a wall, and the thing you end up owning is entirely yours.
The honest downside is time. If the app is the business, building it yourself is the job for a while, and the brief on page one is what stops it from being the job forever.
The cost that outlives the build: running it
Every route ends in the same place. Whoever builds it, the app becomes a service that has to run, a database that has to be backed up, storage, a domain with a certificate, and logs to read when something breaks. The build is a one-off. The running cost arrives every month for as long as the app exists, and it is the number almost nobody asks for in the quote.
So whichever route you choose, insist on two things at the end: the repository is in your account, and the app deploys from that repository onto infrastructure you can log into, with a plan you can read the price of. On RunxBuild that shape is a service deployed from GitHub on the $4 Dev plan or the $6 Basic plan, a managed Postgres or MySQL database on the same ladder beside it, build logs, runtime logs and rollback to the previous deploy. The cost of building the app is a conversation with whoever builds it. The cost of running it should be a number you knew before that conversation started.
How this fits the rest of the stack
The build quote and the running cost are two different numbers, and only one of them shows up unprompted. The RunxBuild hosting calculator gives you the second one before you commission the first: the service, the database, the storage and the bandwidth as separate line items, so that the app you are about to pay for arrives with a monthly bill you have already seen.
Useful related references:
- Next.js MCP Server: Built-in Coding Agent Access to App Internals
- Vercel Database: The Marketplace, the Built-In KV, and the Question the Search Results Do Not Answer
- Free App Creation Software: What You Actually Get, and What You Pay Later
- Services on RunxBuild
FAQ
What is the cheapest way to get an app built?
A no-code or AI builder for the draft, then a freelancer to finish it, is usually the cheapest route to something real. Building it yourself is cheaper in money and more expensive in time. An agency is the most expensive and the most likely to ship a finished product on the first attempt.
How much does it cost to get an app built?
Anywhere from a few hundred to several hundred thousand, depending on route and scope. The number people forget is the running cost afterwards: a small app is typically a service on a plan in the four to twenty dollar range plus a managed database, every month, for as long as it exists.
Can I get an app built without coding?
Yes. No-code builders assemble screens and data in a browser, and AI builders generate the application from a description. Both produce a working draft quickly. The limits show up when the app needs accounts, file storage, integrations or a backend, which is where you either export the code and deploy it, or hire someone.
Should I build a native app or a web app?
For most small business ideas, a web app. It works on every phone, can be added to the home screen, and is one deployable service plus a database rather than two native codebases plus a backend. Go native when you need hardware access or app store distribution specifically.
What should I own when the app is finished?
The repository, in your own account. The hosting, the domain, the database and any app store accounts, in your name. A written note of how to build and deploy it. If a developer or agency is reluctant about any of those, sort it out before the contract, not after.