I built FastAPI AlertEngine while building a payment platform at 2am in Zimbabwe 🇿🇼
It was 2am. I was deep in building a payment orchestration platform, and it hit me — I had no way of knowing my system was degrading until a customer felt it first. Dashboards require you to be watching. Customers don't wait.
The technical problem is straightforward: Most monitoring tools track average latency.
Averages lie. If 5% of your requests take 8 seconds, your average still looks fine. Meanwhile, 1 in 20 customers is hitting a wall. P95 latency tells the real story. So I built a FastAPI middleware that tracks P95 inline — no external agents, no Prometheus exporters, no dashboards to configure.
One line:
instrument(app)
When the health score drops, Claude AI diagnoses the root cause in plain English and sends a WhatsApp alert with a tap-to-approve recovery link. Nothing executes without explicit human approval. Everything is logged immutably.
Here's the honest part: I deliberately packaged it as a standalone product. Getting funding in Zimbabwe is genuinely hard. The payment orchestration platform is the bigger vision. AlertEngine is the vehicle — something I could ship, charge for, and use to fund the build.
It solves a real problem – I needed it myself at 1am. But it also buys me the runway to finish what I actually set out to build.
That felt worth being honest about.
Free SDK: pip install fastapi-alertengine
