End-to-end request flows
These sequences cross several services. Each service page repeats the parts that belong to it in more detail.
1. Sign in and call a GRE service through the gateway
2. From a bet to a recommendation (live path)
3. From a bet to a recommendation (batch path)
4. AI tagging, from page to confirmed tag
5. Tag change reaches recommendations
Tag similarity flows from the Taxonomy API into the recommender only through a catalogue build. Nothing is pushed.
6. Live pipeline view
The Simulation page shows live counters for every service without each browser polling every service.
A service that does not answer within 3 seconds shows as null, and the browser draws that box as offline.