Protecting Corporate Accounts with LiveView realtime dashboard
summary
Cabify replaces a stale Looker Studio + BigQuery prototype with a Phoenix LiveView dashboard built in two weeks by one backend engineer, cutting BigQuery query costs and reducing refresh delays. Journey state flows: message queue → Postgres as source of truth → PG2 pub/sub to propagate updates across clustered nodes so every agent sees changes instantly. ETA values come from a single GenServer, ETA cached and updated every 30s, supporting ~90 agents and ~50k weekly journeys and hitting a 99% assignment success rate.
AI-written summary · about
chapters (7)
- 0:00Real-time dashboard intro
- 3:15Internal tool constraints
- 5:43Dashboard tech options
- 7:06Why Phoenix & LiveView
- 8:31Dashboard data flow
- 11:22Scalability decisions
- 13:28Lessons & recommendations
related talks
30mTales of Using Phoenix LiveView at an Early Stage Startup
40mThe Power & Performance of Phoenix LiveView
28mLobbying with Elixir: Real-Time LiveView and OTP
35mKeynote: Fireside Chat on LiveView, (...)
46mThe joy of building your startup with Phoenix LiveView
24mBuilding an Interview Scheduler in LiveView: A Newcomer's Journey.
01h 13mKeynote: Phoenix LiveView - Interactive Apps without Javascript
34mIntroducing LiveData: LiveView for JSON endpoints