Since you are posting this in a Developer Forum, the goal is to make it clear that the AI’s failure isn’t a user error, but a systemic technical failure that has caused financial and operational damage to your business.
Use this description. It is professionally aggressive, technically specific, and leaves no room for them to blame “bad prompting.”
TOPIC / SUBJECT:
URGENT: Material Regression & Credit Exhaustion — 30-Day Service Failure on SmartSign AI
DESCRIPTION:
Account Level: Google AI Pro / Antigravity Ultra
Project Status: 95% Completion (Pre-Migration) → 80% Completion (Post-AG)
Total Credit Loss: 100% (Exhausted on non-functional loops)
I am filing this formal notice of Service Failure and demanding an immediate full refund. For over 30 days, I have attempted to finalize the production launch of SmartSign AI. Instead of completing the final 5% of development, Antigravity has systematically destroyed a working codebase, resulting in a catastrophic regression of my product.
The Failure Breakdown:
-
Total Disregard for
MASTER_BLUEPRINT.MD: Despite providing a comprehensive architectural blueprint and explicit “Launch Checklists,” the agent repeatedly ignored core constraints. It replaced stable, multi-tenant logic with hallucinated “spaghetti code” that it could not then debug. -
Destruction of Working Auth/Checkout: My product had fully functional Supabase Auth (Login/Reset) and Polar.sh integrations prior to using this tool. Antigravity has since broken the authentication flow and stripped out environment variables and scripts for the checkout, rendering the live site (
smartsignsai.com) unusable. -
Algorithmic “Credit Bleeding”: On four separate occasions, the agent entered an infinite “Audit-Fix-Fail” loop while trying to implement a simple Broker Dashboard update. It consumed my entire monthly credit quota without delivering a single line of functional code.
-
Database/RLS Failure: The tool is incapable of correctly managing Supabase Row Level Security (RLS). It failed to configure the necessary policies for Profiles and Notifications, effectively locking me out of my own application data while burning credits to “try” again.
I have spent a full month auditing and attempting to guide this tool, only to have it break my business-ready product.
This is not a “learning curve” issue; this is a Defective Service issue. The orchestrator is failing to maintain context and is actively causing material damage to user projects.
DEMANDED RESOLUTION:
-
Immediate Full Refund for the last 30 days of subscription.
-
Manual Credit Restoration to allow for the manual cleanup of the damage caused by the agent.
-
Escalation to the Engineering team regarding the regression of imported stable code.
I have documented the “Code Red” downtime this has caused for my company. I expect a response from a human representative—not an automated bot—within 24 hours.
POST TAGS:
#bug #feedback #rate-limits #regression #billing-failure