Skip to content
S SiPesan Docs
ID EN

Conversation Summary

Conversation Summary shows the gist of a chat right in the agent workspace, so whoever is handling it — especially an agent who just inherited the chat — grasps the context immediately without re-reading the whole history. The summary is generated automatically by AI from the conversation itself.

What the agent sees

The Conversation Summary card appears in the right-hand info panel, above the status card. It has two main parts:

  • Core questions — the customer's main need or questions, briefly summarized.
  • Answered — the answers already given, each with who gave them (the AI, the chatbot, or the previous agent's name).

For a customer who has chatted before, an extra Previous interactions section lists that customer's past sessions, each with its date, whether it was resolved, and who handled it — so the agent can spot an open loop before replying.

The card can be collapsed but never disappears — just click its header to reopen it. A refresh button is there for when an agent has been chatting a while and wants the latest version of the summary.

When the summary appears

The summary refreshes automatically at every hand-off point:

  • when the AI hands the conversation to a human agent (escalation),
  • when a chat is transferred from one agent to another,
  • when the session closes (saving the final state to read later).

Because it is stored, the summary shows up instantly when a returning customer opens a new chat — no waiting for reprocessing. When a chat has just been handed to an agent, the card auto-expands so it isn't missed.

Turning it on

This feature is off by default and is switched on by the SIPESAN team for your business. Once enabled, the card simply appears on every chat with no extra setup on your side.

One requirement: the summary is generated by AI, so your business needs a configured AI provider — a cloud API key is enough, without having to switch on the AI Agent to answer customers. If no provider is set, the card still appears but shows "summary not ready yet"; summary generation is best-effort and never interferes with the chat itself.

Limits & privacy

Summaries are drawn only from conversations within your own business — never data from another business. The agent-facing summary is kept separate from the memory the AI uses when answering customers, so the two don't cross over.

To configure how the AI answers customers (rather than just summarizing), see the AI Agent guide. For the big picture of the agent workspace, see the all-services overview.