Frontend and full-stack
TypeScript everywhere, and a framework that renders on the server unless the product is purely an application behind a login.
We default to
- TypeScript
- React
- Next.js
- Astro
- Tailwind CSS
- Vite
About
AI coding tools have made building a working application genuinely easy, and have changed nothing about what it takes to run one. PutThrough exists in that gap.
01Studio
The name is the standard: if we build it, you can put real customers, real payments and real traffic through it.
PutThrough is a production engineering studio. We take applications built with Lovable, Bolt, Cursor, Replit and the tools like them, and we do the work that stands between a convincing demonstration and a service people can pay for: access control, payment correctness, deployment, monitoring, recovery, and getting through App Store and Play Store review.
That work is unglamorous and it is largely the same work from one app to the next, which is why we sell it at a published price instead of quoting it. We build new things too — web applications, mobile apps and AI agent features — but those are quoted, because a new build has no typical size and a published price would only mean padding it.
We are not the right studio for everyone. We do not design products, we do not do growth, and we do not take on a rescue we do not believe we can finish. When the honest answer after an audit is that an application should be rebuilt rather than repaired, the report says that — even though the sprint is the thing we would rather sell you.
02Commitments
Six commitments. They are on this page so you can hold us to them, and so you know what the engagement feels like before you are inside one.
An audit needs viewer access and nothing more. We do not ask for production credentials, write access or customer data to tell you what is wrong. Write access is requested when a sprint begins and revoked the day it ends.
The scope is written down and agreed before work starts, and the price does not move. If we find more than we estimated, that is our problem — we carry the estimation risk, because we are the one who did the estimating.
Written reports, written handovers and recorded walkthroughs, so progress does not wait for a meeting to exist. We hold a daily overlap window for calls and we keep them short.
Everything happens in branches and pull requests you review, on infrastructure you own and bill. There is no layer of ours left behind that you would need us to maintain, and no lock-in to walk away from.
Every change is described in plain language in a handover document. The test of an engagement is whether the next engineer can pick it up without us, and that document is how they do it.
Both are available and signed before access is granted. If your customers are in the EU or UK, the data processing agreement is the one you will need for your own records.
03Location
Engineering is based in India (IST), working async with teams in the US and Europe, with daily overlap for calls.
In practice this is an advantage rather than something to work around: you describe a problem at the end of your day and there is progress on it when you start the next. We do not pretend to be somewhere we are not, and we do not ask you to take a call at midnight.
Calls weekdays 18:30–22:30 IST
Outside these we are async — anything urgent goes to email and is picked up at the start of our day.
04Stack
If your application is outside this, tell us on the call. We would rather say no than learn a stack on your budget.
TypeScript everywhere, and a framework that renders on the server unless the product is purely an application behind a login.
We default to
Boring, typed and synchronous until something measurably needs to be a queue. Most systems need fewer moving parts than they are given.
We default to
Postgres unless there is a specific reason otherwise, migrations in the repository, and a restore that has been executed rather than merely configured.
We default to
Expo unless the product is the platform experience. One codebase is worth a great deal when the maintenance bill outlasts the project.
We default to
Managed platforms until scale or cost argues otherwise. The parts we insist on are a deploy you can roll back and an error that reaches a human.
We default to
Hosted models behind an interface you can swap, with the eval set deciding which one you use this quarter rather than a contract.
We default to
Fixed price, two business days, read-only access. It is also the cheapest way to find out whether you want to work with us at all.
Prefer to write? Send your requirements
Analytics
We’d like to use Microsoft Clarity to see how people use this site — clicks, scrolling and which pages help — so we can make it better. Anything typed into a form is masked. It sets cookies only if you allow it. Privacy policy