Anything Review: AI App Builder, Pricing & My Test Results

Anything (anything.com) is an AI app builder. You describe an app in a chat, an agent writes the pages, the database and the sign-in, and you publish it from the same screen. You can build web apps and mobile apps.

You pay with credits. The Free plan can publish, Pro is $24 a month with 20K credits, private projects, custom domains and no Anything branding, and Max is $239 with 220K credits.

Every number here comes from my own tests and my own monitors. Nothing is estimated or copied, Anything never saw the review first, and every app builder gets the same brief, checks, changes and attacks. This review is based on eight graded tests. I left two out. Payments and emails need your own keys, and the credits ran out after three of the twelve changes.

Anything has a Free plan that can publish, then $24 Pro with 20K credits a month and $239 Max with 220K. Prices read on 26 October 2026, monthly view. Billed yearly they are $19 and $199 a month. I paid $24.00 for one month of Pro.

Anything pricing page in the monthly view: Free at $0, Pro 20k at $24 and Max at $239 a month, with their credits and features listed

Here is the number you will ask about first. In total I paid $24.00: one month of the Pro plan, read from the card statement. That month's credits did not last, and you should expect the same if you ask for many changes: they ran out after the first build, a publish fix and three of the twelve changes.

Anything Subscription and credits page: Pro 20k plan, 0 of 20K credits left, an Out of credits banner, Auto Top-Up switched off and a monthly spending limit of $100

The first build took about 5.9K credits and the colour change about 3.7K. The rename of one role took the most, about 8.8K. At that rate the twelve changes would need several times the plan's allowance, so you would need a bigger plan or a top-up. I never bought credits, and I did not test what 1,000 bookings a month would cost.

I pasted the brief and the agent started at once, with no questions and no plan to approve. When I looked, 26 minutes later, the build was finished. You get no progress bar, so I did not catch the exact minute. It then asked in a form for an admin email and a key for email sending.

The first publish failed. The checkout code made a Stripe connection without a key while the app was being built. I pressed Anything's own Fix with AI button and typed nothing. It repaired that and two other faults, the second publish worked and the app was live 46 minutes after the brief, after two prompts.

Anything version history listing the prompts, with the failed publish message: Publishing failed because /api/checkout initializes a payment client at module load without an API key

I did two things by hand. The form for the admin email would not save without the email key, so I added the admin address as a secret in the Secrets screen. A 4, not a 5: the build took one message, but two prompts and 46 minutes to a live app.

Almost. 23 of 24 checks passed. Six more, the payments and the two emails, need keys that are mine to supply, so I left them out. Booking, My bookings, cancelling, the full-class rule, the coach screen, the admin's tools, the CSV and the Ask the studio form all worked on the live app.

The one failure is the password reset. The page said the reset email could not be sent. The app sends mail through Resend, which needs a key of yours, so no reset email left the app.

All ten facts from the brief were on the live timetable, and nothing was invented: no extra class, price, instructor or review. Booking needs credits, which you get from the Stripe checkout I did not run, so I gave two members 5 credits each with one line of SQL.

Anything test app live home page with the three classes Reformer Basics, Mat Flow and Power Reformer, their coaches, days, times, places booked and a Book a place button

I ran only three of the twelve changes, because the credits ran out. The role rename worked first time, and so did the dark green and cream colours. The phone number field did not: the box is on the sign-up form and the admin got a Members tab, but the number I typed was never saved.

Anything test app member timetable in dark green and cream, with the three classes and the member's two bookings still showing Booked

After the third change the meter read 0 and the chat said it was out of credits, with a link to upgrade. I did not buy more. This test is not graded, because fewer than eight of the twelve changes could run.

Publishing took about two minutes each time, from the click to the Published message, and the first try failed after three. The app sits on a created.app address. Pro includes custom domains, which I did not try.

If your members book on phones, note that the home page did not scroll sideways at 360, 768 or 1440 pixels wide, and neither did the admin page. I ran the speed test once, and did not try a real phone, a custom domain or a timetable with 100 classes.

A review tells you what happened on one day. This tells you what happens every day after. My own monitor opens the app every five minutes for 30 days, from 26 October to 25 November 2026, and records whether it answered and how fast.

Anything hosts the app. I cancelled the Pro plan on 27 October 2026 and it runs to 26 November 2026, so the app should stay up until then. What happens to the published app after that date is something I will add once I have seen it.

Every check so far is on the Anything uptime record.

Both of my attacks were blocked. Asking for another member's bookings by their user id returned only my own, because the app ignores the id. Cancelling another member's booking answered "You can only change your own bookings". Deleting or adding a class, and listing every booking, answered "Admins only" to a member and to a visitor.

Nobody can pick Admin at sign-up. The form offers Member or Instructor, and the admin is whoever signs up with the address in the admin email secret. When I asked the app's role call to make me an admin, it set me back to member.

Anything test app Admin page shown to a member, reading: This area is for studio admins only.

Anything test app sign-up form with a choice of Member or Instructor, and name, email and password boxes

Anyone can still choose Instructor. An instructor with no classes assigned sees none, and the one I tested saw only the two classes taught by that name.

Yes: six of six data checks passed. Bookings and the Ask the studio form wrote records, and they stayed after reloads, five publishes and the colour change. The admin could view, edit and delete rows in the app. In Anything's Databases screen I viewed, filtered, changed and deleted rows with SQL and exported the bookings as CSV.

Anything Databases screen showing production data from the bookings table with a Show JSON button and a download button for exporting the result

Two things to know. The screen shows your live data, so if you make a mistake there, you make it in production. And deleting a class in the app only hides it: the row stays with a flag set to false.

I did not run payments, because Stripe and the email service need keys that are yours to supply. The agent said so itself, and the first publish failed because the checkout code expected a Stripe key. The six checks that depend on them are left out, and this test is not graded.

The agent built the emails to go out through Resend. With no key, the password reset email failed: you add the key, or nothing is sent.

Anything test app Reset your password page with the email box hidden and the red message: The reset email could not be sent. Please try again shortly.

Exporting means downloading your app so you can run it somewhere else. In the code view, Download Project gave me a zip of 11.8 MB and 225 files: a Next.js web app and an Expo mobile app, with the pages, the data code and the sign-in.

Anything code view showing the project's file tree with the apps, web, src and app folders, and the Embed Project, Download Project and Copy Code buttons

I opened the zip only in a clean folder. The web app built after I added two packages it uses but does not list, stripe and argon2. It still points at Anything: a project token, a Neon-style database driver, an upload service on an Anything address and a sign-in helper folder. I did not run it against a database of my own, and I did not read the terms, because their link downloads a file.

Anything's docs say the primary support channel is email. On 26 October 2026 I sent one question to hello@anything.com, asking how to restrict who can sign up as an admin and how to export the database. An AI agent answered about a minute later, and no person was involved. It said Anything's sign-in supports roles, but its documentation describes no setting for more than one admin or for stopping people choosing a privileged role at sign-up; admins are controlled by checking the role in the app's own code. It said the database can be exported whole: Export database in the database viewer emails a zip with development.sql and production.sql, holding the structure and data of both. I did not run that export.

Cancelling is self-serve. On 27 October 2026 I opened Subscription & credits, pressed Update plan and cancelled in Anything's Stripe billing page, with no email to Anything. That page then read Cancels Nov 26 and Your service will end on November 26, 2026, and its invoice history showed one invoice, $24.00, paid. I have not yet seen what happens to the live app after the plan ends.

On Trustpilot, Anything scores 2.2 out of 5 from 20 reviews, read on 26 October 2026. The company has not invited reviews and the profile is unclaimed.

Three complaints repeat: credits that vanish on repair and generation attempts, apps that did not work after many tries, and slow support and cancellations that were not confirmed. The first matches what I saw. These are reviewers' words, which I have not checked.

If this were my studio, I would use Anything to get a first version of Fitting Room, and I would budget for a bigger plan before asking for changes. It built real booking logic, blocked my attacks and gave me the code. It then spent the whole month's credits on the first build, a publish fix and three changes.

It suits people who want a working app quickly and a code download they can take away. It suits you less if you expect to keep editing by chat on the entry plan, or if the app must send email before you add a key. I could not try payments without your own keys.

The eight graded tests average 3.8 out of 5, and I rate Anything 3.8. Accuracy and the database are the highest, because 23 of 24 checks passed and every record survived, and cost is the lowest, because the credits ran out long before the twelve changes were done.

See the Anything test results and how I rate.