How do human agents run many conversations at once?

One person, many conversations

In a traditional queue, a person owns a conversation from start to finish and can hold three or four at once. In BotableX the bot owns the conversation. People are pulled in for the moments that need them, then step back out. The same person can stand behind far more conversations, because most of the time they are reading and steering rather than typing.

Live Conversations

Live Conversations is the team’s view. It lists every open conversation across web chat, email, WhatsApp and Messenger, with the channel, the customer, the bot handling it, the current state and how long it has been waiting. Each conversation is marked Bot when the bot is talking, Needs help when it has asked for a consultation, and With agent when a person has taken over. Nothing is hidden; a manager can open any conversation and watch it happen.

The view opens in Tabs or Split layout. Split shows the Customer Chat on one side and the Bot Consultation on the other, so you can read what the customer sees and talk to the bot privately at the same time.

Conversation States

Conversation States are buckets you define to sort the list the way your team thinks: “New”, “Waiting on customer”, “Needs a manager”, “Refund review”, or whatever fits. Each state can have an auto-move trigger, so a conversation moves itself when it is the bot’s turn, when the bot has replied and is waiting on the customer, when the bot has asked for help, when a person has taken over, or when the conversation closes. If HubSpot is connected, a state can also map to a pipeline stage so the ticket moves too.

SLA countdowns

Each conversation shows how much time remains against your response targets, with separate targets for urgent, high and normal priority. The list marks each one as On track, Due soon or Urgent, and a breached target is flagged so it stands out. Sorting by remaining time gives the team a simple rule: work from the top.

Routing help requests

When the bot asks for help, the consultation is routed to the least busy teammate who is Available, within a maximum number of concurrent conversations you set per person, between 0 and 10. A teammate who is at their cap is skipped. Someone marked Busy or Offline receives nothing. A load badge next to each name shows the team how much everyone is carrying.

Guidance versus takeover

Most consultations end in one line of guidance: the teammate reads the context, types what the bot should say or check, and the bot turns it into a reply in your voice. The teammate is free again in seconds.

For the conversations that need a person, Takeover puts them in the customer chat directly and silences the bot for that conversation. Release to AI hands it back with full context. Let Bot Continue closes a consultation without taking over when the bot has already found its way.

Marking conversations for learning

While a conversation is open, a teammate can mark it Always learn or Don’t learn and add a training note. Those marks feed the nightly learning run, so the team’s judgement improves the bot without anyone editing a prompt.