Caller memory

Caller memory lets your agents remember the people they talk to. When someone calls back, messages on WhatsApp, or opens your web chat again, the agent already knows who they are and what happened last time. It greets them by name, confirms what it knows instead of asking again, and carries on from where the last conversation stopped.

How it works

1. The caller gets in touch by phone, WhatsApp, or your web chat.
2. The agent reads their memory first. Before it says a word, it knows their name, what they asked for, what was promised, and what happened last time, even if that was on a different channel.
3. The memory is updated afterwards. When the conversation ends, what is new is added: a changed address replaces the old one, a new order is recorded, a finished task is cleared. Small talk is left out.

An example

Rahul applies for a job and is screened by your recruitment agent over the phone. The screening starts but doesn't finish. The next day he messages on WhatsApp, and your WhatsApp assistant replies "Hey Rahul! We spoke about the opening. How can we help?"

When he calls again, the agent opens with what it remembers and offers to pick up the screening where it stopped.

How to think about it

• It is a notebook, not a script. The memory is a short record of a person: a summary, key facts, open items, and a few things to avoid. Your agent's instructions still decide what it does; the memory only tells it who it is talking to.
• The caller is always right. If a caller corrects something, the agent believes them, and the memory is updated after the conversation.
• Agents remember in teams. You group agents into Memory pools. Agents on the same pool share one memory of each caller; agents on different pools never see each other's.
• The latest conversation wins. If someone chats on WhatsApp and then calls, the agent follows the call. What happened most recently is what counts.
• Nothing is remembered by default. An agent that is not on a pool remembers nobody.

Next steps