Custom web app
Built by a developer or an agency
A custom web app on React, Next.js, Node or another modern stack. One bug, one fix, no retainer.
Bug fixing serviceWeb app rescue: we fix bugs in live web apps
Worked yesterday, broken today? Login, payments, data, email, a deploy or a custom domain stopped working in a custom web app, a Lovable, Bolt or v0 app, or a Zap. We trace the error to the piece that broke, then give you the evidence and a fixed or capped repair price before we touch production.
Tell us what broke in a short form: about three minutes, free to send. The $99 diagnosis is credited toward any repair you approve, and nothing in production changes without your written approval.
Choose your fix: custom web app, Lovable app or Zapier automation
AI-built app
Built with an AI builder
Lovable, Bolt, v0, Replit or Cursor: a vibe-coded app that reached production.
Lovable app fixAutomation
Built on Zapier
A published Zap stopped or produced the wrong result. One failed run, one fix.
Zapier fix serviceNot sure
Not sure which one? Describe what broke
Skip the reading. Tell us what broke and we'll take it from there.
Describe the incidentWhy a live web app breaks after the developer has moved on
A modern web app leans on a stack of moving parts — database, auth, payments, email, APIs, deployment, DNS. Whoever wired them together has moved on, and nobody owns them when one changes or quietly expires.
That's the gap we fill: a fixed-scope diagnosis, decisions in plain language, and no production change until you approve the scope and the price.
The symptom is simple.
The cause may be somewhere else.
Is your problem a fit? Previously working features only
-
Something that worked is broken now
Sign-in, payments, data, email, deploy or a domain that used to work has stopped.
Start an Incident Diagnosis -
Outside what this desk takes on
New builds, redesigns and anything outside our stack are a different job.
See what we do not take on
Never worked correctly in the first place? Tell us anyway — we say whether it fits before you pay.
Web app failures we fix
Login, Supabase, Stripe, deploys, email, DNS, secrets and integrations that worked before and stopped.
Auth
Login and authentication
Users unable to log in, signup failures, password reset issues or a broken 2FA flow, usually right after a provider or config change.
Database
Supabase and database issues
Permission errors, 403s, policy regressions, schema changes and failed migrations.
Payments
Stripe and payments
Checkout errors and payments failing, a customer who paid but has no access, or webhooks not updating orders.
Deploys
Deployment and blank screens
Fine before the last deploy, broken after it: build failures, hydration errors, environment mismatches.
Email and APIs
Email and external APIs
Emails not sending, API errors, webhook failures and other third-party integration issues.
Bug fixing service for web appsDNS and SSL
Domains, DNS and SSL
Custom domains not working, SSL errors, DNS misconfigurations.
Config
Secrets and environment config
Missing or invalid secrets, wrong environment variables, key rotation.
Integrations
Broken integrations
Connected services, CRMs and other external integrations that stopped syncing.
Who we help: abandoned, inherited and AI-built apps
Your developer or agency is gone.
The app was built by a developer who has since left, and nobody picked it up. We work from the code and the connected services, so we don't need whoever wrote them.
You inherited the codebase.
It runs, you own it, and nobody has read it end to end. Code written by another developer is the usual starting point here.
There is no one to call when production breaks.
Nobody is on call, nobody owns it, and the failure is costing you now. For teams without in-house devs, an existing app still needs someone to answer when it fails.
You've tried fixing it with AI.
Each prompt changed something, the first bug is still there, and now there are others.
Start a $99 Incident Diagnosis
You approve the scope and the price before anything changes.
How a bug fix works, from intake to a verified repair
Simple. Human. Effective.-
Send a short bug fix request
Tell us what worked before, what fails now, and what changed, in a short form of about three minutes. Nothing is charged when you send it.
-
A human investigates
We establish the last-known-good state and trace the failure across the relevant systems.
-
You receive a diagnosis and decision
The evidence, our confidence, the affected component, the risk, and what the repair will cost.
-
We restore and verify
Only after approval: the smallest safe change, tested end to end, with a record of what changed.
Bug fix pricing: $99 diagnosis, then a fixed or capped repair
One incident,one price.
Incident Diagnosis
$99
Covers one incident in one app and the systems directly connected to it. You get the evidence, the risk and a fixed or capped repair proposal.
Credited toward an approved repair.
Standard Restore
$249–$499
A contained, reversible and verifiable repair after diagnosis — configuration, credential, permission, deployment and bounded integration failures.
Diagnosis credit applies.
Complex Rescue
$750–$1,500+
Multi-service failures, custom code, data repair, migration impact or higher-risk production changes.
Written scope, cap, rollback plan and acceptance criteria before work begins.
These are indicative ranges, not automatic quotes. You approve the written scope and price before any repair work begins. Third-party platform and usage fees are not included.
Send a bug fix request
Free to send. The $99 diagnosis is credited toward any repair you approve.
Audits, reviews and Lovable migration
Not a broken feature? These are scoped engagements, priced before you pay.
Before launch
Going live with real users soon
Find launch blockers before real users do.
From $749
Guide: It looks 90% done: nine preview gaps Production Readiness AuditSupabase security
Not sure who can read your data
Verify who can see and change what in Supabase.
$399–$499
RLS & Data Isolation ReviewLovable dependency
Deciding whether to leave Lovable
Map and reduce your Lovable dependency before you decide anything.
From $499
Guide: When to migrate away from Lovable: a 5-criterion matrix Production Independence AssessmentMigration
Ready to move off Lovable Cloud
Plan, move and verify a Lovable migration. Planned in stages.
Scoped after a migration plan
Lovable migration serviceScope: what we fix and what we do not take on
Better to say so here than after you've filled in the form.
In scope
What we fix
Web apps built on React, Next.js, TypeScript, Node.js, Supabase, PostgreSQL and Stripe, plus the authentication and deployment around them.
An AI-built app counts too: Lovable, Bolt and v0 ship on Supabase and Stripe, so the same failures apply.
Out of scope
What we don't take on
- WordPress and other CMS, theme and plugin work
- Marketing sites, brochure sites and static websites
- Native iOS and Android apps
- Stacks we do not take on: PHP/Laravel, Rails
- New apps or features, full rewrites, long modernisation, visual redesign
- Maintenance retainers and monthly plans, because this is one-off work
- Hacked-site cleanup, malware removal, penetration testing and live breach response
For a suspected live breach, go to a security-incident specialist and do not send evidence through a public form.
A useful diagnosis, or your $99 back
For a qualified incident, once we have the access and information we need, you get at least one of: a verified or strongly supported root cause, a working limited repair that fits the diagnosis scope, or a concrete external blocker and next-step plan. If none of those land, we review a refund of the $99 under the Terms.
We ask for the least access that will work.
We use secure, least-privilege access and never make production changes without your explicit approval.
Read our access and security processSenior production judgment
Founder-led: 30 years of production engineering
The tools are new. Production discipline is not.
App Rescue Desk is led by its founder, a principal engineer with 30 years on production systems that weren't allowed to go down. AI can build fast. The harder part is knowing which error log is lying, what you must not touch, and when to roll back — then explaining that call to someone who doesn't read stack traces.
That's why the work runs as a desk. One named owner stays accountable for your engagement from first look to sign-off, and the work is written down and gets a second read — so what you depend on doesn't ride on one person's memory.
Alp Bozkurt
Founder & Principal Production Engineer
Frequently asked questions
More questions? Read the full FAQ →What does “previously working” actually mean?
The affected feature successfully completed its intended job before the incident. Last week people could sign in, the payments went through, and the emails arrived.
A feature that has never worked correctly is implementation or development work, not incident restoration - and it is not what this service is for. If you are unsure, complete the intake first. You are only sent to payment if the request matches the incident service.
Do you fix websites, WordPress or online stores?
No. We work on web applications, not sites. WordPress and other CMS, marketing and brochure sites, and native mobile apps are outside the stack — see what's not included. If you are unsure which one you have, describe what broke and we will tell you before you pay.
My developer or agency disappeared. Can you take over the app?
Yes, for one incident at a time. We do not need the person who built it: we read the code and the connected services ourselves, establish what the app did when it last worked, and tell you what broke. Taking over long-term maintenance is a separate conversation — this desk starts with one fix.
What kinds of apps do you support?
We work on web applications built on React, Next.js, TypeScript, Node.js, Supabase, PostgreSQL and Stripe, together with the authentication, email, API, deployment and domain services around them. Apps built with an AI builder such as Lovable, Bolt or v0 sit inside that stack.
How fast will someone look at my incident?
During desk hours a human reviews paid incidents within 15 minutes.
We usually give a first technical read within 30–45 minutes once we have the information and access we need. Actual repair time depends on the cause, application state and third-party services involved.
Do you guarantee a one-hour fix?
No. Some known configuration and integration failures can be repaired quickly, but a universal one-hour repair promise would be misleading.
What we do guarantee is a fast human review and a useful diagnosis for a qualified incident. Any repair estimate comes after the initial investigation, not before it.
What does it cost to fix a production bug, and when do I approve the price?
You start with a $99 incident diagnosis. We explain the likely root cause, the affected component and the proposed repair, and you receive a repair price, fixed or capped, before additional work begins.
The $99 is credited in full toward any repair you approve - for example, if the agreed total repair price is $299, the remaining balance after diagnosis is $200. No production changes are made without your approval.
Will you need my passwords, or change production code straight away?
No to both. Never send passwords, API keys or access tokens by form, email or chat. You grant access through a collaborator invitation, a role-based account, a scoped temporary credential or a supervised screen-sharing session.
We investigate first and explain the proposed action. Production is changed only after you approve the repair scope and price. Full detail is on the access and security page.
Do you offer ongoing maintenance or development?
No. App Rescue Desk is an incident restoration service, not a general development agency. New features, full application builds, redesigns and ongoing development are outside the scope of the service.