RankDude

Replit video review: AI App Builder, Pricing & My Test Results

Miles Foster tests the Replit AI app builder hands-on on a paid Core plan: one written brief for a Pilates booking app, 30 checks, two attacks, twelve changes, publishing, the code export and cancelling. Tested 25 and 26 October 2026; the live screens were recorded again on 26 October 2026. The nine graded tests average 3.7 out of 5, and his rating is 3.7 out of 5.

Review test on film · 8:26
Published
26 October 2026
Length
8:26 min
Belongs to
Replit review

Chapters · 14

This video belongs to the review Replit review, where the grades, the evidence and the written verdict are. Filmed by RankDude during the test; nothing is staged.

The narration is a synthetic voice reading Miles Foster's words. What is on screen is the real test.

Chapters

Press a chapter to play the video from there.

Transcript

0:00 · Intro and the short version

Hey, I'm Miles, and this is RankDude, where I test AI tools hands on and publish every number, so you can decide before you pay. Today it's Replit, an AI app builder. You describe an app in a chat, and its Agent builds the screens, the database and the log-in. I tested it on a paid Core plan, with one brief for a Pilates booking app called Fitting Room. The short version is that one message gave me a working app. The first version passed twenty-eight of my thirty checks, and it blocked both of my attacks. But Replit would not publish it until I removed the payments, and on a phone, the live page is slow. My rating is three point seven out of five. That's the average of nine graded tests. Support isn't graded yet, because Replit hasn't replied.

0:52 · What is Replit?

So, what is Replit? You get a chat box, a live preview, a database tool, and hosting. You say what you need, and the Agent writes the screens, sets up the database and adds sign-in. After that, every change is another message. Replit pays for all of that in credits. The Core plan I bought comes with twenty dollars of credits each month. Emails are not built in. Replit asks you to connect your own email account.

1:17 · One message to a working app: the build

Every builder I review gets the same brief, word for word. Fitting Room is a booking and membership app for a small Pilates studio. It has a class timetable, members who book, coaches who see only their own classes, an admin, and a ninety-five dollar five-class pass. I sent the brief in Power mode. The Agent worked for about seven minutes, plus three short turns. It stopped thirty-seven seconds in, to ask me to connect Stripe. It asked twice for my email key, which I declined, and it asked which email should be the first admin. So the wall clock ran much longer. The first version passed twenty-eight checks about an hour after my message, but it could not be published while it used Stripe. It went public only after a third message removed payments, eleven hours and thirteen minutes after the brief. Most of that was waiting for a decision, not the Agent working. On the published scale, that's a two.

2:13 · Accuracy: 30 checks

Now accuracy. I ran thirty checks on the first version, in a private preview, each with a screenshot. They cover sign-up, log-in, roles, the timetable, booking, cancelling, a full class, the coach's view, the admin's tools, the contact form, payments, and the emails. Twenty-eight of thirty passed. The two that failed are the emails. The app told me it had saved the booking and that delivery was pending, which is honest, but no email arrived. All ten facts from my brief were correct. But the app invented one rule. It says credits come back if you cancel before class starts, and I never asked for that. Accuracy scores four.

2:55 · Log-in and two attacks

Log-in, and the two attacks. Sign-up with an emailed code worked, so did logging in and out, and so did a password reset by code. Attack one. I'm member A, and I ask for member B's bookings. Blocked. Asking for another member's booking returned only my own rows, and trying to change or delete one was refused. Attack two. A member calls the admin actions. Refused too. The coach sees only the classes they teach. I did not run Replit's own security scan. Five out of five.

3:26 · Database and admin view

Next, the data. The app saved bookings, orders and form messages as real records, and they were still there after twelve changes. The admin can see and change them, and download all bookings as a spreadsheet file. Replit's Database tool shows a separate development database and a production database, each with its tables and row counts. After I published, production held my test records: bookings, orders, classes, inquiries, users and the waiting list. Five out of five.

3:56 · Payments and emails

Payments and emails. In the first version, Replit set up the Stripe test connection itself, and payments worked. A ninety-five dollar pass and a twenty-four dollar single class both went through. The orders showed for the admin, and the credits were right. A card that is meant to fail was refused, with no order and no credits. The emails did not arrive. Replit needs your own email account and key, and I gave it none. The app showed its emails as pending, and said so. I then removed payments, so the grade rests on the first version's checks. Four out of five.

4:31 · Twelve changes

Then I asked for twelve changes, one message each. Things like a phone number field, a renamed role, new colours, an About page, a coach filter, a waiting list, a dark mode, and a bug I planted on purpose. Ten worked in one message. One could not be finished, because the second email needs your own email key. And for one, the phone layout, I did not record the result. One of the ten changed nothing, because the rule was already true. Each change took one to five minutes of the Agent's work. I ran a short check after each one, and found nothing broken. Four out of five.

5:07 · Publishing and phone speed

Publishing. Replit would not publish the app while it used a Stripe test connection. The Publish button stayed greyed out, and claiming the test account led to a request for a live Stripe account. I stopped there. I had the Agent remove payments, deleted the sandbox connection by hand, and pressed Publish. The app answered at its public address about seven minutes later, with no log-in wall. On speed, Google's phone test scored the home page forty-four, forty-five and forty-five. The main content took about eight seconds to appear. I did not try my own domain, because I own none for the test. That earns a two.

5:45 · Can you take the code out? Export

Can you take the app with you? On Core, yes. The project downloads as one file, with nine hundred and ten files in it, thirteen point one megabytes. But it won't run alone. It needs your own Postgres database, your own sign-in keys and a Stripe key. I unpacked and read the code, and I did not run it elsewhere. I also did not read Replit's terms, so I can't tell you who owns what you build. Read them first. This test scores three.

6:12 · Support and cancelling

Support and cancelling. Support is a Get help form inside the app. It warns that Core requests go to the standard queue, and Pro requests are answered first. I sent a technical question on the twenty-sixth of October, and Replit has not replied yet, so support isn't graded. I will add it when it does. Cancelling is self-serve in billing. A confirmation lists what you lose, then a Stripe page asks for a reason. I cancelled on the twenty-sixth of October. The service ends on the twenty-fifth of November, and my app stays live until then.

6:46 · What it cost

What did it cost? In total I paid twenty-one dollars seventy-eight, which is one month of Core plus tax. Building and changing the app used about five dollars of the twenty in credits. The first working version cost about two dollars, and the twelve changes about two dollars eighty. The last time I read the usage page, fourteen dollars ninety-nine was left. I removed Stripe and published afterwards, and I did not read the usage page again. So those costs, and the cost of keeping the app running, are not in these numbers.

7:17 · Pricing

Pricing. Replit has a free Starter plan, with daily agent credits and one live project. Core is twenty dollars a month, with twenty dollars of credits. Pro is a hundred, with up to fifteen collaborators and ten parallel agents. Enterprise is custom. Paying for a year is cheaper. Building and changing my app used about a quarter of the Core credits, so Core was enough to build. Pricing scores four.

7:42 · Verdict: 3.7 out of 5

So, my verdict. Replit built a working booking app from one message, passed twenty-eight of thirty checks, blocked both attacks, and got ten of twelve changes right in one message. The weak points are that it would not publish with Stripe connected, the emails need your own key, and the live page scores forty-five on a phone. If this were my studio, I would try Replit for a simple booking app, and skip payments at first. It suits founders who want a booking or admin app fast. It suits less if you need payments live on day one. The nine graded tests average three point seven out of five, and that is my rating. Every measurement, every screenshot and every price is in the written review and the test results, linked below.

The written review

Every grade in the video is explained on the review page, with the screenshots and the test results.

Replit review

More videos on ai app builders