Feb 2026 · 7 min read · EPAS · Case study

Designing a Conversation Console for Real Customer-Care Work

A practical note on conversation history, message templates, delivery states, filters, pagination, and the backend contracts behind a Zalo OA operations module.

#Zalo OA #Angular #Operations #REST #Redis

Problem Context

A customer-care console has to answer operational questions quickly: which conversation needs attention, what was sent, which template was used, what state the delivery reached, and how to move through history without losing context.

Architecture & Design

The module combines a conversation-oriented UI with explicit backend contracts for filtering, pagination, template management, status tracking, and cacheable active-session data. The important design decision is to model the operator's work context instead of presenting a collection of unrelated API tables.

Key Technical Decisions

  1. 1Keep filters and pagination explicit so operators can reason about which slice of history they are viewing.
  2. 2Separate template administration from conversation history while keeping preview and send-state feedback close to the operator's task.
  3. 3Use cache selectively for active conversation context without treating cache as the source of message truth.
  4. 4Prefer clear empty, loading, and failed-delivery states over pretending that the conversation is always real-time.

Key Engineering Takeaways

  • Operational UI quality is mostly about preserving context through state changes.
  • A message console should communicate uncertainty and delivery state honestly.
  • The best abstraction starts from the operator's next action.