Flow Lab

AI-built app: customer data protected and verified

This React and Supabase booking demo passed a security and logic check. After the fixes, customer data stays private. The full report shows how each fix was checked.

This demo is self-initiated, not a client project. The business and the first version of its app are fictional. The situations in the demo are typical of Supabase apps made with AI builders. Every number about the demo comes from the demo itself, so you can check each one yourself.

Screenshot: 10 fixes made to an app built with an AI builder
10fixes, 4 of them top priority
16 of 16test attacks blocked after the fixes
20 of 20normal actions still work
1 business dayto get a price and timeline for checking your app
Case study

The project goal. A fictional dog-walking business built its booking app in an AI builder, using React and Supabase. The app already worked in a quick walkthrough. The task was to test everything a customer does in the app.

Results of the check. The app got 10 fixes, and all of them now work in the app. Customer addresses and phone numbers are protected by access rules. The master database key was removed from the browser code. Each walker takes only one booking per time slot, so there are no double bookings. Booking times sort correctly, so 9:00 always comes before 10:00. The owner sees each day's schedule in order.

16 of 16test attacks blocked after the fixes

Verifying the fixes. Each of the 7 database fixes was tested with a real attack, run once before the fix and once after it. The other 3 fixes went into the browser code and were checked by reading it. All 20 everyday actions still work, so customers use the app just as before. Before the database fix updates past bookings, it checks each one. If a booking needs attention, the update stops and leaves the database unchanged. The owner gets a list of those bookings, so they stay under the owner's control.

Testing ran on PostgreSQL 18.3 with Supabase's default access rules. The interface was checked by reading the code.

Your order includes a check of data access, security and everything customers do in your Supabase app. You then get a report showing what can be improved in the app and what each improvement costs. After the fixes, the same checks run again, so every change is confirmed. AI drafts the code in Claude Code, Anthropic's AI coding tool. Flow Lab runs the attacks and checks. You grant code and database access for the check and close it after handover. The work never needs your account passwords, so they stay with you. I quote the price and timeline for the check once you describe your project. I quote each fix from the report. Each fix gets a set price before work starts. You pay in stages, each one after you approve it. All amounts are in US dollars. You can pay through Contra or by PayPal invoice. Work starts 2 business days after you confirm the quote. A quote holds for 7 days. Optional support costs $150 a month. It covers up to 6 hours of work. Each request gets a reply within one business day. New changes pass the same checks. You can cancel support month to month.

Next step

The same check for your app.

Send a link to your app. Add one line saying which builder made it. Within one business day, you receive a plan, a price and a timeline. The review is free. Any bug found within 30 days of delivery is fixed at no charge.