Tell it what you need in your own words. A small team of AI helpers plans it, builds it and checks it — on your computer, on the AI account you already pay for.
Free during the beta. No card, no sign-up fee.
Sound familiar?
Chat-style AI is good at answering. It is much worse at finishing — and that gap is where your afternoon goes.
You list five things. It does two, and says it is done. You only notice the other three when you go looking.
The same AI that wrote it also grades it, and it always gives itself a pass. You find out it was wrong much later.
Some tools edit your folder while you are still reading. If something breaks after that, finding it is your problem.
What you can do
None of these need you to write a line of code. You describe the job, you read what came back, and you decide whether to keep it.
SPREADSHEETS
“Every month I get three sales files from three branches. Merge them, drop the duplicate rows, and give me one sheet with a total per product.”
WHAT COMES BACK
You still have to open the result and check a few rows yourself. It does the boring part, not the trusting part.
A SMALL INTERNAL TOOL
“Our team books meeting rooms in a group chat and it is a mess. Make a simple page where you pick a room and a time, and it lists who booked what.”
WHAT COMES BACK
Good for a tool your own team uses. Anything customers touch needs a developer to look it over first.
TIDYING UP FILES
“This folder has 2,000 photos with names like IMG_0423. Rename them by the date they were taken and sort them into folders by month.”
WHAT COMES BACK
Keep a backup of anything irreplaceable anyway. That is true of every tool, not only this one.
THE UNFINISHED PROJECT
“I started this eight months ago and forgot how it works. Read it, tell me where I stopped, and finish the sign-up screen.”
WHAT COMES BACK
This is where it shines. Picking up cold work is exactly the part people never get back to.
CODE SOMEONE ELSE WROTE
“A contractor left us this. Explain what it does in normal words, then change the company name and the logo everywhere it appears.”
WHAT COMES BACK
It reads it all from your own disk — and the one doing the reading is the AI account you connected, not a server of ours.
THE WEEKLY REPORT
“Every Monday I copy numbers out of four files into the same slide deck. Do that for me and leave the layout alone.”
WHAT COMES BACK
Best when the files arrive in the same shape every week. If the shape keeps changing, so will the fixing.
And what it is not. This is not “an app on the App Store by tonight”. What it is good at is the small, repetitive, annoying work that sits between you and going home — the kind nobody has time to build properly.
A job like the ones above usually takes anywhere from ten minutes to an afternoon, and part of that time is you reading and saying yes.
How it works
Type it the way you would say it to a colleague. There is no special wording to learn and nothing to set up first.
One helper decides the order, one does the work, and a third one checks it and sends it back if it falls short.
You see exactly what will change, in full, before it happens. Approve it or throw it away — both take one click.
The team
They are not three chat windows. They hand work to each other, and the last one is allowed to reject the work of the one before it.
Turns one sentence from you into a list of steps in the right order, and says which files it expects to touch. You can rewrite that list, or cut it down, before anything starts.
Actually opens the files, writes the change and runs it to see whether it works. Everything it writes goes into a copy, so your own files sit untouched the whole time.
Reads the finished work against what you originally asked for. If a piece is missing it goes back to the Builder, and you never see that round trip.
Why bother splitting it up? Because one AI marking its own homework is the reason you get half a job. Here the checking is done by a different helper, with your original request in hand and no attachment to the work.
The honest answer
You do not have to write anything. But you cannot walk away either — somebody has to look at the result and say yes, and that somebody is you.
YOUR PART
ITS PART
So: no, it is not fully automatic. Anyone telling you their AI works with nobody watching is selling you something. The useful question is not “can I stop paying attention” — it is “how little attention does this need”. Here, it needs you at the start and at the end, and not much in between.
If you have never opened a code editor in your life, that is fine. Start with the chat-style app further down this page.
Safety
Every change is written into a copy first. Your own files sit exactly as you left them until the moment you press approve.
The old version and the new one, line by line, with the changed parts marked. Nothing is summarised away.
Right before anything is applied, the state of your folder is saved. One click puts it all back the way it was.
Five settings, from asking before every single move to working through a long job while you are somewhere else.
Three things are blocked at every setting. It cannot reach outside the folder you opened. It cannot edit or delete the file that holds its own rules. And anything you have written down as forbidden stays forbidden, right up to the most hands-off setting. Those three are not a preference you can switch off.
One exception, said out loud. If you connect a subscription like Claude Code instead of a key, that company's own program does the editing, in its own window — so on that route there is no preview before the change, only the roll-back point afterwards. The app tells you this every single time before it starts.
Your machine, your files
Your files are read and written on your own disk. They are never uploaded to us, because there is no server of ours for them to pass through — we never built one.
The only thing that leaves your machine is the request itself, plus the parts of the file it needs to read — and that goes straight to the AI company you chose, not to us.
Because nothing passes through us, the only question left is your own company's rule about sending anything to an outside AI service. Ask before you point it at work material.
The thinking happens at the AI company, so you need a connection — unless you run a model on your own machine, which it also supports.
The money
Monoprise is the app around the AI, not the AI itself. You bring the account. Nothing you spend on it comes to us, and we take no cut of it.
If you pay monthly for Claude Code, OpenAI Codex or Google Antigravity, sign in once and keep using the plan you already buy. Nothing extra to pay.
Paste a key from Anthropic, OpenAI, Google or Groq and they bill you directly, by usage. There is no plan to buy, and you can stop any time.
Nothing. The beta is free, and there is no card to enter. We will charge for the app itself one day, and we will say so clearly before we do.
One caution, so it does not surprise you. If you pay by usage rather than by plan, a team of three helpers reads and writes more than a single chat window does, so it costs more per job. The app shows you the running cost, and you can cap it or switch to a cheaper model for the routine parts.
Two apps
They do the same work with the same team and the same approval step. The only difference is how much of the machinery you see.
One window that looks like any messaging app. You say what you want, and what comes back is written in normal sentences, not code you have to decipher.
A full code editor with the team living in a side panel. If you already use VS Code, this is that, with your settings and extensions still in place.
You can switch later without losing anything. Both apps share the same settings and the same AI accounts on your machine.
Where we are
We haven't launched. Below is what works today and what doesn't, with nothing quietly left out.
And the plain truth: it still gets things wrong. It misreads what you meant, it takes a longer route than a person would, and sometimes it is simply confident and mistaken. That is exactly why nothing is applied without you looking at it first.
Questions
You never have to write any. But you do have to read what came back and decide whether it is right — nobody else can do that for you. If you have never opened a code editor, use Monoprise Chat and you will not have to see any code at all.
Leave your email at the bottom of this page and we send you the build within a day. You unzip it and double-click. It does not need admin rights and it does not install itself into Windows, so deleting the folder removes it completely.
The app is free during the beta and there is no card to enter. What you pay for is the AI itself, and you pay that to the AI company, not to us — either through a plan you already have or by usage on your own key. We add nothing on top.
They stay on your disk and we never see them. There is no Monoprise server in the path — we never built one. What does leave your machine is your request and the parts of your files the AI needs to read, and that goes directly to the AI company you chose, under their terms.
Technically yes, since nothing passes through us. The real question is your employer's policy on sending anything to an outside AI service, which is the same question as using ChatGPT at work. If your company runs its own model in-house, you can point Monoprise at that instead, and then nothing leaves the building.
Yes, for the AI part — the thinking happens at the AI company's end. The only exception is a model running on your own computer, which works with no connection at all but needs a fairly powerful machine.
Changes are written into a copy first, so until you approve, your own files have not been touched. And right before anything is applied, the state of your folder is saved, so one click puts it all back. Keep a backup of anything irreplaceable anyway.
Anything vague. “Make it better” gets you nothing useful. It is also weak on work that depends on knowledge only you have — your company's rules, why a file is shaped the way it is, which of two correct answers your boss wants. And it is not the tool for building something big that customers will pay for.
A small tidy-up job is minutes. Something like the monthly report example can take an afternoon, mostly because you will go back and forth a few times before it is right. It is not instant, and anyone promising instant is not counting the part where you check the result.
If you do not write code, Monoprise Chat. It is one window, it explains itself in normal sentences, and you never have to look at code unless you ask to. Pick Monoprise IDE if you already use VS Code and want the team next to your editor.
Yes. There is a stop button, and it stops the whole team, not just the sentence being written. Whatever had already been written into the copy is still there for you to look at, approve or discard.
It is kept on your own machine and used to talk to the AI company directly. It is never sent to us, and it is hidden from the logs the app keeps. If you signed in with a plan instead of a key, that sign-in stays where the AI company's own tool put it.
Something we didn't answer? The developer page goes into the machinery →
Leave your email and we'll send you the build within a day. Tell us what you're hoping to make and we'll write back about whether it's a fit.
Beta · free · no card
Prefer to just email us? monoprise@monoprise.dev
Used only to send you the beta build. No newsletter, nobody else sees it.