VEITH.DEV
AI-Built Apps · Production Hardening

You Vibe-Coded It. We Make It Production.

Your AI-built app works, and real people use it. Now it needs the parts no prompt writes: tests, pipelines, security, an owner. We install them on your existing code. No rewrite, no lecture.

No rewrite by default  ·  Fixed-price audit  ·  Stuttgart, EU

Why Does Every Change Break Something Else?

AI tools are astonishing at writing code and indifferent to running it. They ship the demo; they don't carry the pager. What a vibe-coded app is missing is never talent. It's the engineering system around the code.

No Safety Net

There are no tests, so every change is a gamble. You find out in production, from a user, on a weekend.

The Loop

The codebase outgrew the context window. Your AI tool rewrites the same file back and forth, and the bug survives every round.

Open Doors

API keys in the client, database rules wide open, auth that was never reviewed. Fine at 10 users. A liability at 1,000.

Nobody Owns It

When it goes down at 2am, who is accountable? A subscription is not an engineer. Prompts don't carry pagers.

To be clear: the app is not the problem. It proved your product and found your users. It just never got the boring parts.

What We Install: The Production Layer

On your code, in your repo, around your workflow. Installed in weeks, not quarters.

01

Production Audit

Fixed price, 5 days. One repository, one environment, read access to what runs today. Anything larger we requote before we start. We read the whole codebase: security, data model, dependencies, infrastructure. You get a written report and a ranked plan, whether you continue with us or not.

$ audit ./your-app
read: 142 files · next.js + supabase · you, cursor, claude
CRITICAL service key shipped in the client bundle
HIGH row level security off on 4 tables
HIGH payment flow: zero test coverage
# full report: every finding ranked, each with a fix
02

The Safety Net

A test suite around the behavior your users depend on, and a CI/CD pipeline that runs it on every change. Staging before production, rollback in one step. Deploys stop being events.

03

Security & GDPR

Secrets out of the code, auth hardened, database rules locked down. EU hosting, records of processing, an AVV when your customers ask for one. Production-grade means GDPR-grade here.

04

Guardrails for Your Workflow

Context files so your AI tools know the rules of your codebase, and review gates so nothing unreviewed reaches production. Keep vibe-coding. The system catches what slips.

05

An Owner

A senior engineer who signs the work and stays reachable. And if you want more than hardening, the same team runs your product long-term as your development partner.

We're Not the Cleanup Crew

Most shops that offer this start by telling you your code is garbage and quoting a rewrite. We won't, for a simple reason: we build with AI agents every day ourselves. Vibe coding is not the mistake. Shipping the result without engineering is. The difference between a demo and a company was never talent: it's tests, pipelines, review, ownership. That system is our product.

Keep your stackKeep your workflowGain an owner

The difference: other shops take the keyboard away. We build the system around it.

How It Runs

Audit
fixed price, 5 days
Hardening
fixed scope, weeks
Ownership
optional, ongoing

Start scoped; stay if it works.

Questions, Answered

Do you rewrite vibe-coded apps from scratch?

By default, no. Working software deserves respect: we harden what exists with tests, pipelines, and security fixes. A rewrite only happens when the audit proves it is genuinely cheaper, and then it is your call, with the numbers on the table.

Can I keep building features myself with AI?

Yes. That is the point of the guardrails: we set up the context files and review gates so your AI tools work inside the rules of your codebase and nothing unreviewed reaches production. You keep the speed; the system catches what slips.

What does it cost?

The audit is a fixed price, agreed before we start. The hardening sprint is scoped and quoted from the audit findings, so you know the number before committing. No hourly meter.

Which tools and stacks do you cover?

Apps built with Lovable, Bolt, v0, Replit, Cursor, Claude Code, or any mix of them. Typical stacks: Next.js, React, Supabase, Firebase, Vercel, Postgres. If yours differs, ask. The method transfers.

Is my app a GDPR problem?

If it stores personal data of EU users on default settings, quite possibly. Common findings: no records of processing, US-only hosting, database rules far too permissive. The audit covers all of it, and the fixes are part of the sprint.

Real Users Deserve Real Engineering.

Send us the repo link. You get a fixed-price audit, a ranked list of what would break next, and a plan. Useful whether you fix it with us or without us.

Next Step

You Bring the Problem. We Ship the System.

One call to scope it. If it's a fit, you'll usually see working software within days. No discovery workshop, no deck.