Conversations and Inbox Basics
This guide explains how to work in the Conversations area of Irmaya, where your team reviews customer threads, uses AI assistance, and steps in when human handling is needed.
Overview
The Conversations inbox is the main place to review incoming WhatsApp activity.
In this workspace area, you can:
- monitor new customer messages
- review AI-generated reply suggestions
- take over conversations that need a person
- follow message history and customer context
Before You Start
Make sure these foundations are in place:
- WhatsApp is connected
- bot settings have been configured
- knowledge content has been added
- handoff rules have been reviewed
How Conversations Work
The inbox brings together the customer message, the current thread, and the tools needed to respond.
What you should expect to see:
- recent customer message preview
- conversation history
- AI reply suggestions when available
- handoff or review status
- customer context and contact information
Review the Inbox
Use the inbox to identify what needs attention first.
Helpful review patterns:
- check new or recent conversations
- look at conversations marked for review
- confirm whether a conversation can stay assisted or needs a person
Use Filters
Filters help teams focus on the right conversations.
Depending on the current product flow, useful views may include:
- all conversations
- recent conversations
- unread conversations
- conversations that need review
- flagged conversations
The exact filter labels may evolve, but the main goal stays the same: prioritize the conversations that need action.
Use AI Assistance
AI in the inbox should be treated as assistance, not blind automation.
Use AI suggestions to:
- respond faster to common questions
- keep replies consistent
- pull from the current knowledge setup
Before sending a reply:
- read the suggestion carefully
- edit the wording if needed
- make sure the answer fits the customer’s situation
Handle Human Handoff
Some conversations should be handled by a person instead of the bot.
Typical handoff triggers include:
- the customer asks for a person
- the request is sensitive or complex
- the AI does not have enough confidence
- the issue needs specialist attention
When a conversation needs human support, the operator should:
- open the conversation
- review the full context
- take over the thread
- continue the response as a person
Best Practices
- review suggested replies before sending
- monitor flagged or escalated conversations regularly
- use the thread history before responding
- keep human handoff decisions consistent across the team
- use repeated issues as feedback for bot and knowledge improvements
Common Issues
AI suggestions are missing
Check:
- WhatsApp connection status
- knowledge content availability
- bot configuration quality
Too many conversations need human review
Review:
- handoff rules
- confidence thresholds
- knowledge gaps
- prompt and bot setup quality
Conversation context looks incomplete
Check:
- the current conversation view
- history loading
- whether filters are hiding older context
Related Guides
- Product Concepts — Understand core product areas
- Bot Development — Improve assistant behavior
- Human Handoff Support — Review handoff logic
- Quick Start Guide — Set up the workspace
- Troubleshooting — Resolve common issues