ALL
POSTS
41 posts so far.
Our AI Rate Limiter Decided 0 Requests Per Second Was the Correct Limit at 2 AM
We built an AI-powered adaptive rate limiter that monitored traffic patterns and adjusted limits automatically. At 2:07 AM on a Saturday, it analyzed an anomalous traffic spike, classified it as a DDoS attack, and set our public API rate limit to 0 requests per second. It was our own mobile app doing a scheduled sync.
We Let Claude Write Our Database Migrations — Here's What It Silently Deleted
We used Claude to clean up deprecated columns from our users table. The migration ran in 180ms. What we didn't know: a billing service in a separate repo was still reading those columns. 3,200 invoices went out with blank shipping addresses before anyone noticed. Recovery took 14 hours.
We Asked GPT-4 to Review Our Pull Requests for 30 Days — It Approved the Bug That Took Down Prod
We integrated GPT-4 as an automated code reviewer into our GitHub Actions pipeline. Over 30 days it caught 61 real issues. Then it approved a subtle auth bypass that left 23 admin routes unprotected for 11 days — because the bug lived across three files that were never in the diff.
We Set temperature=0 and GPT-4 Still Gave Different Answers — Our Entire CI Pipeline Broke
We built an automated code review pipeline that used GPT-4 with temperature=0 to enforce consistent output. After OpenAI silently updated the model behind the same API endpoint, our determinism assumption collapsed — tests started flipping between pass and fail on identical inputs, and we couldn't reproduce failures locally.
The AI-Generated Migration That Dropped the Wrong Index and Took Our API from 3ms to 45 Seconds
We asked an AI coding assistant to clean up redundant PostgreSQL indexes. It dropped the wrong one. API latency jumped from 3ms to 45 seconds. Here is what a 3.5-hour investigation taught us about trusting AI with schema changes.
Our OpenAI Bill Went From $23 to $4,200 in 48 Hours — A Missing Stop Sequence Did It
We built a feedback-processing pipeline that used GPT-4 to categorise and summarise user feedback. A single missing stop sequence caused the model to loop indefinitely, generating 40-million tokens of circular output over a holiday weekend while our alerts stayed quiet.
Our AI Documentation Bot Invented 14 API Routes That Never Existed — 6,000 Users Integrated Against Them
We shipped an LLM-powered documentation assistant trained on our API docs. Within three weeks, it had confidently hallucinated 14 non-existent endpoints. Developers built integrations against them. Support tickets arrived. We had to choose between breaking those integrations or actually building the routes the AI had promised.
The AI Agent That Cleaned Up Our K8s Manifests and Crashed Production
We let a Cursor AI agent refactor our Kubernetes deployment files to remove boilerplate. Six hours later, 34% of requests were failing as pods OOMKilled faster than they could restart.
PAGE 1 / 2 · 9 POSTS