Impact:
100% ML parameters reliance (dropping the manual settings from MGMT)
ML predictions precalculated (BE)Note and not in real time (FE)
Allow Chatbot feature on all tasks
Main Changes:
Now | After | |
Task’s input/output param | 🧑🔧 Defined in MGMT on action level | 🧠 Predicted by the Brain |
Default zones/tools | 🧑🔧 Partial predefined in MGMT | 🧠 Predicted by the Brain |
Dashboard<>ML integration | Some are FE<>ML and some BE<>ML | Always BE<>ML (consistent) |
Task info step loading time | ± 5 seconds (real time FE<>ML api call) | Seamless (precalculated by BE) |
Chatbot (app) | Triggers only for categories (e.g send gifts) | Triggers for ALL tasks |
Tech changes:
The task creation would happen immediately
All the other processes will be in the background
ML flows will return the response based on the current request state
Anything related to the actions is out
New implementation as microservices
Some of our collections will not be used after this change
Triggers for ticketing services (creation, matching, suggestion, assignment) would happen from the new request management service
Ticket: