For apps built with Lovable, Bolt, Replit, v0 or Base44

Your app has real users. Now it's breaking.

Slow pages, a half-built feature, failing logins, data you're not sure is safe. We find what's wrong, rank it, and fix it in your account.

Book a slot and give written permission. We look at your app before it; in the slot, we walk you through what we found.

Every step, one relationship

  1. Safety check

    Free

    With the owner's written permission, a look at the biggest risks (exposed keys, open databases, broken logins) beforehand, then a 30-minute call to walk you through what we found

    Book a free check

  2. Rescue audit

    US$499 once

    A written report of every problem ranked by severity, a verdict for each part (patch, rework or rebuild), and a fixed quote for the repair list: a price for each repair and a total

    Request an invoice

  3. Fixed-price repair

    The audit's quoted price

    The audit's whole repair list as one job: one price, one finish date, a re-test of every repair, and a before-and-after note

    Request an invoice

  4. Care

    US$99 a month

    Monitoring, backups (including one restore test when the plan starts), a weekly security and dependency check, a monthly report, and up to one hour of small fixes

    Request an invoice

  5. Ship

    US$999 a month

    One request at a time, about three working days each: security fixes, bug fixes and small features, every change tested and reviewed

    Request an invoice

  6. Build

    US$2,500 a month

    Priority work, larger features, integrations and payments, and moving the app onto its own infrastructure when it outgrows the AI builder

    Request an invoice

Every step includes

  • You own the code, in your own account.
  • Every change ships with tests and a plain-English note of what changed.
  • Monthly plans have no long contracts; pause or cancel any time.
  • The fixed-price repair: one price, agreed before work starts. Anything not on the list is priced separately.

How it works

  1. Book a free check

    Book a 30-minute slot and give written permission. Before it, we look at your app's biggest risks in our own time; in the slot we walk you through what we found, in plain words.
  2. Get the audit

    A written report of every problem, ranked by severity, with a verdict for each part and a fixed quote for the fixes.
  3. Fix it all at once, or over time

    Fix it all at once: order the fixed-price repair, the whole list as one job, with one price, one finish date and a re-test. Or fix it over time: start Ship or Build straight away, and the audit's list becomes your queue, worked through one request at a time, most severe first, with no finish date for the list and no re-test of the whole list. Either way, anything exposing your users' data now goes first.
  4. Choose Care, Ship or Build

    One monthly plan for what comes next, in your own account, with tests and a plain-English note on every change. Pause or cancel any time.

Your app's database

Many apps built with AI app-builders keep their data in Supabase, a hosted PostgreSQL database. It is where the worst problems hide: tables anyone can read, backups nobody has tested, and no clear way out when you want to move. We secure it, with access rules on every table tested with two accounts; set up backups and prove a restore works, in a temporary copy inside your own account; and move it to your own hosting when you outgrow the builder. It runs through the same steps above: the audit finds the problems. The fixed-price repair makes the fixes the audit finds, or Ship works through them one at a time if you'd rather spread the cost. Care keeps watch afterwards and looks after the backups, and Build moves it to your own hosting.

See a sample report

A complete rescue audit of a made-up app, labelled Sample: every problem ranked by severity, a verdict for each part and a fixed quote, so you can see exactly what you'd get. Read the sample report

Questions

What does the free check look at, and why do you need written permission?
The biggest risks first: exposed keys, open databases, broken logins. We never scan or test an app without its owner's written permission, so the booking form asks for it. You get the findings in your 30-minute slot.
What's in the audit report?
Every problem we found, ranked by severity, with a verdict for each part — patch, rework or rebuild — and a fixed quote for the repair list: a price for each repair and a total.
After the audit, do I have to fix everything at once?
No, you choose. Fix it all at once: the fixed-price repair does the whole list as one job, with one price, one finish date and a re-test. Or fix it over time: start Ship or Build, and the list becomes your queue, most severe first, with no finish date for the list and no re-test of the whole list. Either way, anything exposing your users' data now goes first. If you decide not to fix a critical finding, we record that in writing.
What does "one request at a time" mean, and how long does one take?
You keep a queue; we work on the top item until it ships, tested and reviewed, then the next one. A request is work that fits in about three working days; bigger ones we split with you or quote separately.
Can I pause or cancel a monthly plan?
Yes, at any time, with no notice period. Pause and you are not charged for the months you're paused; cancel and everything stays yours, in your own account.
Who owns the code?
You do, from the first day. Everything lives in your own accounts — code, hosting, databases — and we work inside them.
What don't you take on?
Apps whose owner won't put permission in writing, and work that would need your users' personal data copied onto our own systems or given to the AI.
How do you use AI?
We do the work with Claude, an AI from Anthropic, on a business account under Anthropic's commercial terms. Your code passes through Anthropic's systems while we work. Anthropic does not train its models on it, and deletes it within 30 days as standard. Anything its automated safety checks flag may be kept for up to two years. Your users' personal data, your production records and your passwords never go to the AI: work that touches real records runs as code on systems you control. We delete our own copy of your code and our working notes within five working days of finishing.
How do I pay?
By invoice: press «Request an invoice» and we'll email it to you. Prices are in US dollars.
Do you build new products?
Yes, quoted per project. Write to us with what you have in mind.

Ready when you are

The free check needs your written permission and a 30-minute slot for the findings.