Live agent conversations
The hard part of running a chatbot is handling the escalations — the moments when the bot rightly says “let me get a human.” On the web, Conversations & live chat is the agent dashboard. On mobile, that experience is coming soon, with a real-time twist.
What this will let you do
- Get pushed when a customer requests a human. → Push: customer requested human
- Take over the conversation with one tap, from anywhere.
- See the bot’s prior turns so you have full context.
- Reply in real time — typing indicators flow back to the customer; the customer’s typing flows to you. This is real conversation, not message-and-wait.
- AI Copilot sidebar suggests replies based on the conversation context (same engine as the seller-side AI Copilot).
- Hand back to the bot when you’re done; the bot picks up where you left off.
- Mark resolved to close the thread.
How it’ll work (preview)
The flow:
- Customer: asks something complex, the bot can’t answer.
- Bot: offers handoff. Customer accepts.
- Push fires to every Hilal Mobile App registered as an agent for that bot.
- First-claim wins — whoever taps first becomes the agent. Others see “claimed by Alice.”
- Real-time chat — Socket.io pipe for live messaging, both directions.
- AI Copilot suggests replies side-by-side.
- Hand back to bot or mark resolved when done.
The Socket.io infrastructure is already wired into the app — see Real-time agent chat for the cross-cutting infrastructure. This page is the chatbot-specific application of it.
Why this matters
Chatbot operators have a structural problem: escalations are unpredictable, and agents aren’t at their desks 24/7. Email-based fallbacks lag by hours. SMS fallbacks lose context. A real-time mobile pipe — push to alert, tap to claim, live two-way chat — closes the gap from “bot escalated” to “human is replying” from hours to seconds.
Status
- Socket.io infrastructure: ✅ wired into the app, minimally used today.
- Live chat UI: the seller-side AI Copilot side panel is reused.
- Bot handoff backend: depends on the chatbot platform’s existing escalation events.
- AI Copilot for chatbot context: model is shared with the seller-side; tuning per chatbot is a small piece of work.
- Beta: ships alongside push for chatbot activity (next major after Notes → KB).